Official Tomato v1.19 released

Discussion in 'Tomato Firmware' started by jeradc, Apr 21, 2008.

  1. jeradc

    jeradc LI Guru Member

    * Removed route hash size changes. This seems to have caused overflow and "rusty's brain broke" errors for some people.
    * Fixed disabled GUI button was not dimmed.
    * Updated DST for France, Germany, Italy, Poland, Sweden timezones.
    * Updated L7 patterns 2008-02-20.
  2. LLigetfa

    LLigetfa LI Guru Member

    There goes my uptime. ;)
    Installed, rebooted, working well.
  3. Kiwi8

    Kiwi8 LI Guru Member

    Wow that's fast!
  4. pdawg17

    pdawg17 LI Guru Member

    How would the overflow problem show itself in "real life"? I haven't noticed any problems with 118...
  5. benny.shen

    benny.shen LI Guru Member

    because the overflow problem only cause error for some people.
    I dont see any problem on my wrt54g.
  6. rickh57

    rickh57 Network Guru Member

    I upgraded my router with the new firmware this morning, too. No problems to report so far.
  7. c007c

    c007c LI Guru Member

    As usual, the upgrade was flawless to my WHR-G125. I have 2 questions.

    1) Do I/should I reset/clear nvram on each new release? I spent a lot of time testing my QoS, so I don't want to clear nvram, or reset, if I don't have to. I have used the same config I started with at v1_17, upgrade only with the last two versions 18 and 19.

    2) If The answer is yes, I should reset/clear nvram for each release, how does the backup/restore work? Does the backup contain only the incremental changes I made? Or is it a complete backup, meaning I would be restoring the configuration back as if I had never reset, putting back old nvram settings not used in this release, etc.

  8. bigclaw

    bigclaw Network Guru Member

    I salute the frequent update of the Tomato firmware. It's great!

    That said, I think I'll sit this one out; it doesn't have any PPPoE fixes. :)
  9. gawd0wns

    gawd0wns Network Guru Member

    Which users are having problems with the increase in hash size? I don't recall seeing anyone posting any problems with 1.18.
  10. nvtweak

    nvtweak LI Guru Member

    According to the FAQ, upgrading to a newer version of Tomato does not require a reset unless otherwise stated in the release notes.

    I personally don't see any reason to move from 1.18 ND, so I will stay with that for now.

    Yes someone on the forum reported having this issue. I imagine Jon probably got some e-mail about it as well.
  11. Kiwi8

    Kiwi8 LI Guru Member

  12. DeCex

    DeCex LI Guru Member

    Nice, fast fix updates. Great as usual. I dont mean to be picky, but the Linux Icon in the "About" page is static, try resize your browzer, see that the icon is sticky, lol.

    Thanks Jon and the community for the great firmware update & infoes.
  13. LLigetfa

    LLigetfa LI Guru Member

    A search of this board or google for tomato "rusty's brain broke" should turn up something.

    If there are no stability issues, I would not bother clearing NV-RAM but if you do have problems, clearing NV-RAM and hand jobbing in the settings are in order before crying "bug".
  14. valerima

    valerima LI Guru Member

    I have tried
    The SpeedMod-patched Tomato 1.19 firmware
    with my Linksys 54 WRT 54 GL 1.1
    speed decreased at once. After reflashing to Jon's 1.19 improvement noted. DL with utorrent increased 8-10 times.
    Procedures were the same. What the reason is?
  15. nvtweak

    nvtweak LI Guru Member

    faster peers second time around? who knows..
  16. hardc0re

    hardc0re Network Guru Member

    There should not be such a big difference in speed, and the difference in speed is not really about the DL/UL rate, its more of the response time. Could it be a configuration issue, or just the unpredictability of torrents?

    You would probably not notice any difference between SpeedMod and normal versions unless you have more than 1000 or 2000 connections, in which case SpeedMod should be more responsive.

    The main differences in SpeedMod are:

    1. IP conntrack core, original version is from Linux 2.4.20, SpeedMod from 2.4.29.
    2. IP route core, original version from Linux 2.4.20, SpeedMod from 2.4.21.
    3. Both route and conntrack core modified to use Murmurhash 2.0 algorithm for their hashtables, which in theory should be a lot faster than the old hash used in Linux 2.4.20, and slightly faster than the jenkins lookup2 hash used in Linux 2.4.21 and newer (including DD-WRT & OpenWRT).
    4. The size of the hashtables in conntrack_core and route have been increased significantly, which will speed up routing: conntrack bucket size examples: DD-WRT: 512, Tomato: 4099, SpeedMod: 16384; route table buckets: DD-WRT and Tomato: 512, SpeedMod: 16384.

    You can see this if you telnet to the router and type "dmesg | grep bucket".

    BTW I am also running Tomato 1.19 SpeedMod on a WRT54GL v1.1, and its fast and stable, with around 2000-3000 connections now. :)
  17. DeCex

    DeCex LI Guru Member

    Hi hardc0re, since im not on Linux planet and only been hearing about linux BSD is very stable. BSD core better or Linux core is Linux core ?. Can BSD core be implement as firmware ? (so blur question, but if i dont ask i prolly dont get an answer)
  18. hardc0re

    hardc0re Network Guru Member

    I think in general, BSD is more stable and maybe faster, but has much less software/hardware support available (unless of course you count Apple's OSX as BSD...). Linux is much more common.

    And as far as I know, there is no usable BSD-based firmware for these Broadcom-based routers, although I wont be surprised if there are people out there trying to make it work. Most of our custom firmware, including Tomato, is modified from the Linksys Linux source code which comes with the drivers for the WiFi, etc.
  19. DeCex

    DeCex LI Guru Member

    ok, cool, littlebit clearer now, thanks
  20. kmindi

    kmindi LI Guru Member

    Also no problems for me and my router.
    Thanks jon.
  21. Rudi1

    Rudi1 Network Guru Member

    This is correct answer!I have some problems with settings,but when I clear NV-RAM everything work like a sharm.Big Thanx to Jon for this update!

  22. Rob650

    Rob650 Addicted to LI Member

    I'm coming from DD-WRT and started using Tomato at 1.17. These releases are really coming fast... whats the average time between releases for Tomato?
  23. butterman

    butterman LI Guru Member

    It varies, but on average about 1 per month for the last year or so.
  24. LLigetfa

    LLigetfa LI Guru Member

    It seems like 1.07 was around forever. Here is the cronology:

    1.04 - Submitted 2007-01-24 11:17
    1.05 - Submitted 2007-03-11 13:09
    1.06 - Submitted 2007-03-18 10:00
    1.07 - Submitted 2007-05-20 20:13
    1.08 - Submitted 2007-09-29 20:52
    1.09 - Submitted 2007-09-30 05:17
    1.10 - Submitted 2007-10-07 20:12
    1.11 - Submitted 2007-10-29 20:52
    1.12 - Submitted 2007-12-04 21:01
    1.13 - Submitted 2007-12-05 06:32
    1.14 - Submitted 2008-01-16 21:26
    1.15 - Submitted 2008-01-27 20:37
    1.16 - Submitted 2008-02-17 17:37
    1.17 - Submitted 2008-02-26 20:55
    1.18 - Submitted 2008-04-13 20:07
    1.19 - Submitted 2008-04-20 17:25
  25. Rob650

    Rob650 Addicted to LI Member

    Thanks very much. Before I learned of Tomato I was using the latest stable release of DD-WRT released sometime in 2006 I believe. Quite a change in pace haha.

    Anyway, no problems here. 1.19 is running great.
  26. mstombs

    mstombs Network Guru Member

    Not sure I will install another even number one - the odds seem to live longer!

    Something was bad for me in 1.18, not Rusty's brain but log full of

    Apr 17 21:44:41 wrt54gs user.warn kernel: ip_conntrack: table full, dropping packet.
    Apr 17 21:44:46 wrt54gs user.warn kernel: NET: 180 messages suppressed.
    Apr 17 21:44:46 wrt54gs user.warn kernel: ip_conntrack: table full, dropping packet.
    Apr 17 21:44:51 wrt54gs user.warn kernel: NET: 180 messages suppressed.
    Apr 17 21:44:51 wrt54gs user.warn kernel: ip_conntrack: table full, dropping packet.
    Apr 17 21:50:04 wrt54gs user.warn kernel: NET: 37 messages suppressed.
    Apr 17 21:50:04 wrt54gs user.warn kernel: ip_conntrack: table full, dropping packet.
    but surprisingly haven't had complaints from users...
  27. Sopmod

    Sopmod Network Guru Member

    Props Jon.

    I love Tomato.

    1.19 is working like a champ!
  28. Talon88

    Talon88 LI Guru Member


    Thanks Jon for the quick "rusty's brain broke" errors fix!

  29. kulmegil

    kulmegil Network Guru Member

    Although my net connection is crappy, there is my ISP's whole 100Mbps MAN network between my router and I-net and I'm interested in getting most out of it.

    So what max throughput I can actually expect with ND - tomato 1.19 (or speedmod) on fast routers... that would be probably WHR-G125 or Asus WL500G Premium. Someone made some tests and can post some results?
  30. szfong

    szfong Network Guru Member

    So, your connection to your ISP is 100Mbps, AND you call it "crappy"??? Whatever....

    Depending on what you doing, # of connections, etc.. WHR-G125, one of the fastest Tomato supported routers will max out at about 65Mbps, and a bit less if your doing heavy p2p. Newer Draft-N gigabit routers may be a better option for you.

    Please check the following link if you want further info, as this is a bit off topic:,com_chart/Itemid,189/chart,120/

  31. Menkatek

    Menkatek Network Guru Member

    1.19-ND working well on a WRT54GS v4.

    I couldn't flash tomato-ND.trx through tftp after many tries. But it worked with the web interface.
  32. asantos

    asantos LI Guru Member

    Request to increase "Static DHCP" to 100 hosts

    Hello, friends, from Spain!!

    This is my request I have send to our admirated Jon:
    (I post here for your comments)

    First, thanks for your wonderfull firmware. It's the best!!
    Since version 1.18, "Wireless Filter" increased to 100 MACs. But "Static DHCP" is still limited to 50. Because I use those options in conjunction, It yould be wonderfull for me if you increase too "Static DHCP" to 100.

    That's my suggestion for you.

  33. Edrikk

    Edrikk Network Guru Member

    ND on WRT54g version 3?


    Has anyone tried the ND version (this or previous) on a WRT54g version 3 (v3) router?

    Curious to see if it would work, and if there is any benefit to be had by updating...

  34. fun.k

    fun.k Addicted to LI Member

    hi all (this is my very first post)

    just thought 2 chime in + confirm that 1.19 worked fine for me too

  35. danix71

    danix71 LI Guru Member

  36. mstombs

    mstombs Network Guru Member

    Because Tomato is flexible and configurable, you could do this now by adding a link to a textfile in the dnsmasq extra config - you would need your own system to manage ip addresses/MAC addresses in a spreadsheet or similar - but would also make re-entering the details after rest to defaults easier!
  37. y2kboy23

    y2kboy23 Network Guru Member

    Does anyone know if this version includes the updates from the original firmware from Linksys (4.71.4)?
  38. hardc0re

    hardc0re Network Guru Member

  39. pharma

    pharma Network Guru Member

  40. hardc0re

    hardc0re Network Guru Member

    No problem. BTW there is a hidden setting in Tomato 1.18 and 1.19, that allows you to switch the QOS qdiscs from SFQ to PFIFO.

    #nvram set qos_pfifo=1
    #nvram commit

    Reboot or restart QoS to make it take effect.

    To see if it changed anything, enter this command:
    # tc -s qdisc show

    Before change:
    qdisc sfq 30: dev vlan1 limit 128p quantum 1518b perturb 10sec
    Sent 341129295 bytes 396854 pkts (dropped 772, overlimits 0)
    backlog 87p
    qdisc sfq 20: dev vlan1 limit 128p quantum 1518b perturb 10sec
    Sent 22744765 bytes 19407 pkts (dropped 0, overlimits 0)
    qdisc sfq 10: dev vlan1 limit 128p quantum 1518b perturb 10sec
    Sent 72658471 bytes 954697 pkts (dropped 0, overlimits 0)
    qdisc htb 1: dev vlan1 r2q 10 default 30 direct_packets_stat 39
    Sent 436565347 bytes 1371000 pkts (dropped 772, overlimits 698391)
    backlog 87p

    After change:
    qdisc pfifo 30: dev vlan1 limit 256p
    Sent 229958 bytes 323 pkts (dropped 0, overlimits 0)
    qdisc pfifo 20: dev vlan1 limit 256p
    Sent 63 bytes 1 pkts (dropped 0, overlimits 0)
    qdisc pfifo 10: dev vlan1 limit 256p
    Sent 27430 bytes 493 pkts (dropped 0, overlimits 0)
    qdisc htb 1: dev vlan1 r2q 10 default 30 direct_packets_stat 66
    Sent 280901 bytes 885 pkts (dropped 0, overlimits 370)

    If you want to change back to SFQ, just unset the nvram variable:

    #nvram unset qos_pfifo
    #nvram commit

    Reboot or restart QOS to make it take effect.

    The reasoning behind this is that PFIFO takes up less CPU usage than SFQ, most people dont need SFQ anyway, and SFQ is not very effective for P2P since P2P connections come from the same IP (the PC running P2P software).
  41. Daggerx

    Daggerx LI Guru Member

    Would this change benefit me greatly for online gaming via the 360?
    If so, where and how do I input these commands?

    1.19 rocks, no problems upgrading. I'm using the speed mod version on a wrt54g v3.

    Thanks in advance.
  42. hardc0re

    hardc0re Network Guru Member

    You should telnet into the router to enter these commands. Well in theory, it might decrease latency / ping times in certain scenarios (if you have P2P running, for example).
  43. apelete

    apelete LI Guru Member

    Is it useful in a scenario where there are several computers (at least 2) connected to the router and they are all (or many of them) running a p2p client ?
  44. apelete

    apelete LI Guru Member

    The "show browser icon" feature seems to be broken in this release.
    If I connect using the router IP it does show up, but if I use the router host name to get an access the favicon doesn't show up...

    Can someone confirm this ? Just clear your browser cache and connect to the router...
  45. Edrikk

    Edrikk Network Guru Member

    Well, since nobody answered, I'll post just so it's cached for search engines / search features.

    I just installed the New Drivers (ND) version of Tomato 1.19 ND on a Linksys WRT54g v3.1, and all seems to be well.

    It could be the placebo affect, but browsing on the laptop feels snappier/faster than version 1.18 (with the older Broadcom drivers) did.

    Will post back if I find something's wrong.
  46. StevenG

    StevenG LI Guru Member

    You know, I feel the same. Maybe not, but I was running 1.13 and just upgraded to 1.19, and I too feel this is snappier than the 1.13 at least.
  47. rcordorica

    rcordorica Network Guru Member

    i just want to put it out there that the ND version isn't always better. on my wrt54gl v1.1 the new drivers cause some of my clients to get an unstable connection with lower throughput.

    On my own computer the ND version cause the signal meter to drop a bar or two, from "Very Good" to "Good" or "Fair". I have flashed several times to see if it really was the firmware, and ND just makes the apparent signal lower on my own computer; maybe its just an interaction with my wireless card and the drivers.

    Upping the power level to compensate doesn't help with ND, and also the wireless survey causes the WLAN to stop functioning with ND. Everything works fine with the regular version.

    Curiously enough however, the first time I used ND (when they were first introduced) everything worked fine and my network was really stable. Who knows why, maybe its just the fact i'm in an apartment complex with tons of APs. Or maybe something changed in tomato to cause the problems with ND.
  48. Odin-60

    Odin-60 LI Guru Member

    This is exactly what I'm experiencing, too -- but I'm using v1.19 with the old
    drivers on a Buffalo WHR-G54S. :rolleyes:
  49. pharma

    pharma Network Guru Member

    I agree! I'm also using the Non-experimental version (v1.19) and notice things seem to be quicker and responsive on my WRT54G's than the previous build, which didn't seem possible since v1.18 was already very fast! :)
  50. Edrikk

    Edrikk Network Guru Member

    Well, you're right...
    So I noticed that with the New Driver version, my WRT54g v3.1's connection to the PS3 appears to be faulty. Meaning the PS3 signs out of the PS Network quite often.

    After going to the "Non ND" version, all is well..
  51. hardc0re

    hardc0re Network Guru Member

    The problem with the SFQ qdisc is that its supposed to implement 'fairness', but its not effective at doing it for P2P traffic, because most of the P2P connections come to/from one or two IPs.

    I think SFQ is only useful if you are *not* using P2P, you have many many PCs behind the router, and you are always hitting the uplink limit.

    That being said, SFQ may work better for some but not others, so its best to just try it and see which is better.
  52. rcordorica

    rcordorica Network Guru Member

    I don't think it would be an issue with CPU usage since sfq is a fast hashing algorithm, however pfifo is even simpler, so maybe if your router is heavily loaded it could help.

    This website has really good info about pfifo and sfq:

    In your example you have dropped packets using the sfq scheduler, but none with pfifo. However, i did notice that your sfq limit size is 128p (packets) while pfifo is set at 256p.

    Also the perturb is set at 10: Each of these buckets should represent a unique flow. Because multiple flows may get hashed to the same bucket, the hashing algorithm is perturbed at configurable intervals so that the unfairness lasts only for a short while. Perturbation may however cause some inadvertent packet reordering to occur.

    I assume those are defaults, but if not there is a lot of room for tweaking. Low latency applications would benefit from smaller limits, but your throughput would take a hit.

    AFAIK, sfq gives each connection "flow" equal time in a round-robin fashion. If you have lots of P2P connections and other mixed traffic, then all connections are given a time to send data inversely proportional to the amount of connections. If you have a low number of flows, then sfq acts just like FIFO.

    pfifo is basically "first in - first out" but the above link says it has three separate FIFO buffers. the first is for low latency and always gets sent out first, and the other two buffers will empty into the first when it has sent all the packets.

    I am inclined to think that pfifo should be the optimal scheduler for low latency since it won't allow p2p data to affect its transmit time. p2p of course will have to wait in the queue longer, so it's throughput would go down. I'm ok with that.

    I am still confused however on how this interacts with the GUI based QoS rules. Do the pfifo and sfq scheduler take into account our rules?

    So much info, I never knew the linux network stack could be configred with such detail.
  53. MacOfTheEast

    MacOfTheEast LI Guru Member

    ANOTHER "smooth as silk" upgrade. Thank you, Jon!

  54. rcordorica

    rcordorica Network Guru Member

    Some more research:

    It seems that tomato by default uses htb as its root and then configures a sfq queue for each level defined in your QoS classifications, i.e. high, medium, and low would be 3 different queues.

    # tc qdisc show
    qdisc sfq 50: dev vlan1 limit 128p quantum 1518b perturb 10sec - LOWEST
    qdisc sfq 40: dev vlan1 limit 128p quantum 1518b perturb 10sec - LOW
    qdisc sfq 30: dev vlan1 limit 128p quantum 1518b perturb 10sec - MEDIUM
    qdisc sfq 20: dev vlan1 limit 128p quantum 1518b perturb 10sec - HIGH
    qdisc sfq 10: dev vlan1 limit 128p quantum 1518b perturb 10sec - HIGHEST
    qdisc htb 1: dev vlan1 r2q 10 default 50 direct_packets_stat 9 - ROOT
    The default sfq limit is 128p, but if we change it to pfifo then the default is 256p. However, when I read the pfifo man page it says:
    And the txqueuelen on the router is:
    # ifconfig | grep txque
               collisions:0 txqueuelen:0 - BR0
               collisions:0 txqueuelen:100 - ETH0
               collisions:0 txqueuelen:100 - ETH1
               collisions:0 txqueuelen:0 - LO
               collisions:0 txqueuelen:0 - VLAN0
               collisions:0 txqueuelen:0 - VLAN1
    Only 100. I would think we want our packet limit for pfifo and sfq to be 100 to fit within our queue.

    Or we could raise the txqueuelen. My goal however is to reduce latency, so smaller limits are better. (Interestingly enough, some people have complained Tomato is slow for them compared to other firmwares. maybe the small txqueuelen is why? It is a common tweak to raise this to get more bandwidth)

    Does anybody know how to pass a limit to the sfq scheduler?

    Now knowing how tomato's default QoS works, I think sfq would probably be better for latency since it has smaller default packet limits. Also, since each class has its own sfq queue, P2P shouldn't effect your higher classes which have their own smaller queue.

    My p2p queue at least is at the bottom, or "lowest". So my HTTP and other traffic should hit the first or second queue and get classified and sent before my lowest class.

    Therefore even though sfq is "fair" for an entire queue, its not being fair globally on all your traffic. This should avoid any problems with p2p creating too many flows for sfq to deal with.

    Maybe a solution is to define the p2p class as pfifo, and leave the rest sfq. or the reverse? In any case, I think tomato's default behavior is damn good.


    I ran ifconfig eth0 txqueuelen 128 and ifconfig eth1 txqueuelen 128 and it accepted the changes ok. I don't know what, if any effect it will have. But I think it makes sense to follow the man page's advice and have a matching txqueuelen and packet limit.


    I found some evidence that txqueuelen only matters for traffic that is not being classified:

    Does anybody know more about this?
  55. cpgbg

    cpgbg LI Guru Member

    I have Buffalo WHR-HP-G54 and I have tested the maximum throughput with iperf of 3 different versions of Tomato. I have connected 1 PC to the WAN port and other to the 1st LAN port. I have configured static IP and tested the throughput in both directions. Here are the results:

    Tomato 1.19.2129 SpeedMod:

    download: [1920] 0.0-60.0 sec 421 MBytes 58.8 Mbits/sec
    upload: [1816] 0.0-60.0 sec 417 MBytes 58.3 Mbits/sec

    Tomato 1.19.8427 SpeedMod:

    download: [1920] 0.0-60.0 sec 402 MBytes 56.2 Mbits/sec
    upload: [1816] 0.0-60.0 sec 396 MBytes 55.3 Mbits/sec

    Tomato 1.19.1463 (official release):

    download: [1920] 0.0-60.0 sec 429 MBytes 59.9 Mbits/sec
    upload: [1816] 0.0-60.0 sec 418 MBytes 58.4 Mbits/sec

    It is really strange that SpeedMod is actually slower than the official release, esp. the latest build.
  56. nvtweak

    nvtweak LI Guru Member

    Rodney said it's not so much about the throughput, but rather increasing responsiveness when the router is dealing with lots of connections.
  57. hardc0re

    hardc0re Network Guru Member

    Thanks for benchmarking, those are interesting results. SpeedMod is really tailored to perform faster when there are a large number of connections, and not so much for maximum throughput of a single TCP stream.

    In SpeedMod, some components of the Linux kernel were updated to their latest versions, and the major network hashing algos were changed, partly for security reasons (prevent hash exploits the Linux 2.4.20 is vulnerable to) but also just to keep updated. Maybe the changes in the Linux kernel or the hashes has somehow affected the maximum LAN-WAN throughput if using a single TCP connection.

    I did get some feedback that the cpu usage of the router is actually lower with SpeedMod vs official release under load though. Still, interesting results.

    Maybe you can benchmark with the multiple streams option, and try going to several hundred or (ideally) several thousand (2000-8000) streams if its possible? Not sure if that'll hang iperf or not.
  58. hardc0re

    hardc0re Network Guru Member

    SFQ's limit is hardcoded at 128p. The PFIFO limit in Tomato is set at 256p from my recommendation, mainly to prevent unnecessary packet drops during bursts. It wont increase latency unless there really is a queue of packets building up.

    Personally I prefer PFIFO because it uses minimal CPU and feels more 'snappy' than SFQ for heavy web surfing. I find SFQ's features unnecessary unless you really have a lot of real non-p2p users, AND your uplink is always saturated.
  59. Harry100

    Harry100 Addicted to LI Member

    Had some troubles with my router rebooted automatic with 1.17. Like every 3-4 days. Now I have uptime 6 days, 05:31:24 since the 1.19 update. Looking good. Router WRT54G version 2
  60. cpgbg

    cpgbg LI Guru Member

    I have also tested with 100 concurrent connections with iperf. This tool is not very good at testing with many connections, as you will see that although I have specified 60 sec it took longer and I am not sure that the result it correct.

    Here are the results with 100 connections at once:

    Tomato 1.19.8427 SpeedMod:
    [ ID] Interval Transfer Bandwidth
    [1696] 0.0-60.3 sec 3.03 MBytes 422 Kbits/sec
    [960] 0.0-60.0 sec 1.27 MBytes 177 Kbits/sec
    [944] 0.0-60.2 sec 2.02 MBytes 281 Kbits/sec
    [800] 0.0-69.5 sec 1.61 MBytes 194 Kbits/sec
    [1184] 0.0-69.7 sec 1.30 MBytes 157 Kbits/sec
    [576] 0.0-69.5 sec 2.05 MBytes 247 Kbits/sec
    [SUM] 0.0-69.9 sec 296 MBytes 35.5 Mbits/sec

    Tomato 1.19.1463 (official):
    [ ID] Interval Transfer Bandwidth
    [1792] 0.0-60.1 sec 3.15 MBytes 439 Kbits/sec
    [896] 0.0-60.0 sec 3.02 MBytes 422 Kbits/sec
    [1824] 0.0-60.3 sec 3.22 MBytes 448 Kbits/sec
    [1760] 0.0-82.0 sec 1.77 MBytes 181 Kbits/sec
    [816] 0.0-83.9 sec 72.0 KBytes 7.03 Kbits/sec
    [608] 0.0-83.9 sec 72.0 KBytes 7.03 Kbits/sec
    [SUM] 0.0-84.1 sec 313 MBytes 31.3 Mbits/sec

    The total data transferred of official tomato is more, but somehow there was a big delay while listing the results and I don't trust the numbers.

    The first SpeedMod of 1.19 has almost the same speed as official Tomato for single connection, but the latest build is slower.
  61. StevenG

    StevenG LI Guru Member

    Me too. 1.19 has also been up 6 days. Looking good!
  62. szfong

    szfong Network Guru Member

    Automatic rebooting can sometimes be prevented by flashing with the "ND" release.

  63. schlichte

    schlichte LI Guru Member


    no problem with 1.19

    1.19 has been up since 16 days without any problems.

    Router: wrt54gl


  64. drelkata

    drelkata LI Guru Member

    failed to access /etc/resolv.dnsmasq: No such file or directory

    This message is in my log . I run Tomato 1.19 on wrt54gl v1.1, on other with Tomato_1.16.1374VPN-SDMMC is :
    Purko cron.notice crond[94]: crond 1.9.1 started, log level 9
    Jan 1 02:00:13 Purko dnsmasq[104]: started, version 2.41 cachesize 150
    Jan 1 02:00:13 Purko dnsmasq[104]: compile time options: no-IPv6 GNU-getopt no-RTC no-ISC-leasefile no-DBus no-I18N no-TFTP
    Jan 1 02:00:13 Purko dnsmasq[104]: DHCP, IP range --, lease time 1d
    Jan 1 02:00:13 Purko dnsmasq[104]: reading /etc/resolv.dnsmasq
    Jan 1 02:00:13 Purko dnsmasq[104]: using nameserver
    Jan 1 02:00:13 Purko dnsmasq[104]: using nameserver
    Jan 1 02:00:13 Purko dnsmasq[104]: using nameserver
    Jan 1 02:00:13 Purko dnsmasq[104]: read /etc/hosts - 0 addresses
    Jan 1 02:00:13 Purko dnsmasq[104]: read /etc/hosts.dnsmasq - 5 addresses
    Jan 1 02:00:13 Purko daemon.crit dnsmasq[89]: failed to bind DHCP server socket: Address already in use
    Jan 1 02:00:13 Purko daemon.crit dnsmasq[89]: FAILED to start up

    Any ideas ?!?!
  65. JensG

    JensG Network Guru Member

    I am experiencing some stability problems with my Linksys WRT54GS v1.0, using Tomato v1.19.1463.
    Twice it has re-booted, and today it stopped responding in periods after 6 days uptime. I re-booted it, and it seems to work again.

    I have been using Tomato since nov or dec 2006, and has never experienced anything like this in the previous versions. My record was more than 70 days with one of the early versions.
  66. JensG

    JensG Network Guru Member

    I can add, that sometimes when I watch the real-time bandwidth graph, it comes with a 10 second timeout, like on the screenshot in this link:
  67. danix71

    danix71 LI Guru Member

    Installed on a GL and working with it as a client. EXCELLENT! GOOD JOB INDEED! :wave:
  68. danix71

    danix71 LI Guru Member

    Please advice: what should be the script for a WRT54GL+Tomato, client to another WRT (router), to light (white) the SES button when a connection exists? Based on ''ping''-ing the gateway, I don't know...suggestions?
  69. JensG

    JensG Network Guru Member

    I think my stability problems got solved by resetting the NVRAM and entering all my settings again. I hadn't done that for ages.
  70. pfoomer

    pfoomer LI Guru Member

    Look Here
  71. danix71

    danix71 LI Guru Member

    pfoomer, thank you very much!
  72. Sunspark

    Sunspark LI Guru Member

    I wonder about the dns caching sometimes.. I have a bunch of DNS servers defined with a 2k cache, and I activated all-servers option too so as to go with the fastest one everytime, and this helped, but still, once in awhile I get a long delay of 'looking up google' or whatever and I can't tell if it's the OS using SRV queries, the web browser, or the router itself.
  73. StevenG

    StevenG LI Guru Member

    24 days of uptime since my upgrade to 1.19 and no problems. Rock solid wireless too. I think I'll be sticking with this version for a long time to come. Nice work Jon! Thanks.
  74. jsmiddleton4

    jsmiddleton4 Network Guru Member

    I'm having page load lags/stalls with 1.19 as well as victek's mod of 1.19. Gone back to 1.17 and no problems, everything is snappy. Have a Buffalo 54HP for primary router.

    Especially on with 1.19. Lots of ads/banners that load is seperate windows. With 1.19 it was almost unbearable. Checked with the forum and their end was fine. Could be me and something coincedental of course as Cox is having issues. But the 1.17 firmware does not seem to have any problems.
  75. rcordorica

    rcordorica Network Guru Member

    I suggest using pfifo. It really improved qos on my wrt54gl. after doing further research, pfifo is the most responsive, and it uses less cpu (hardc0re already pointed that out). I was also getting "Rusty's brain broke!" errors, and only when I removed my l7 filters did they go away. pfifo really helps if you have a lot of connections. I had ~5000 connections for p2p and everything remained responsive (as long as http is classified higher than your p2p connections). sfq (default) doesn't work as well for lots of p2p connections.

    Another thing that causes stalls for me weak wireless signal, which happens at different times of the day (i don't know if you are on wireless).
  76. jsmiddleton4

    jsmiddleton4 Network Guru Member

    I'll look at it. Is this another one that needs to be done via telnet or can we pop it in the script section and try it easily? Not that telnet is all that hard. Its just a lot easier to clear in the script section if you don't want to use it. Also I'm not using qos. Does that matter?

    Right now it seems Cox is partly responsible. Something going on with a server out here in the west.
  77. jsmiddleton4

    jsmiddleton4 Network Guru Member

    Well it seems to be working by putting it in the script area in Admin. It is most definetly faster. AVS Forum loads as quick as I can ever remember it. I mean snaps right up there.

    I'll have to read back and see what it is I'm actually doing......
  78. rcordorica

    rcordorica Network Guru Member

    you only have to set qos_pfifo once. Then it gets saved to nvram. i ssh into my router, but the admin script should work fine.
  79. jsmiddleton4

    jsmiddleton4 Network Guru Member

    Since I was in the thing via telnet anyway to setup the wan to lan thing I went ahead and did the qos thing. Hard to measure the difference but pages most certainly load faster.
  80. hardc0re

    hardc0re Network Guru Member

    The commands to switch QOS qdiscs from sfq to pfifo need to be done only once. Just telnet in, enter commands, nvram commit, reboot. You don't need to put it in the startup script.

    Pfifo will definitely be faster than SFQ for large numbers of connections (for example, P2P usage). From what I can see in the source code of the SFQ qdisc, each qdisc only has about 1000 flow / connection queues, so it isn't going to be effective if you have many more connections than 1000 per qdisc level (normally, 1 qdisc level = 1 QOS priority level).

    Also if you like even snappier response even with many connections, try my SpeedMod version here. The latest build is pretty much optimized now. :)
  81. Sunspark

    Sunspark LI Guru Member

    I solved the slow dns problem awhile ago. It was PEBKAC. Annoying.

    For whatever reason, probably due to testing something and forgot to undo it, I had a DNS server hardcoded in the OS.. ignoring the router entirely.. I've since removed that, and enabled all-servers, and ever since I haven't had a single DNS delay.
  82. jsmiddleton4

    jsmiddleton4 Network Guru Member


    Thanks for the suggestions. I'm using victek's mod and if I'm not mistaken it is using your mods as well. I'm not using a system that has a ton of connections and while we do some p2p, we don't do it all the time. Even at that the qos tweak is clearly faster. No way to measure it in terms of user perception but it is clearly loading pages faster, snappier.
  83. Daggerx

    Daggerx LI Guru Member

    Is the new release for tomato due out soon? Just wondering.
  84. StevenG

    StevenG LI Guru Member

    1.19 has been so solid and stable, it will take a lot of new features to get me to move off this version. 40+ days of uptime and counting.
  85. pumpkin

    pumpkin LI Guru Member

    I think 1.19 is the best release of Tomato til now. Router is up since 35 Days without any problems.
  86. wahur1

    wahur1 LI Guru Member

    using 1.15 and its been up for 56days :D..
  87. Toastman

    Toastman Super Moderator Staff Member Member

    Latency down

    Just thought you'd like to hear that on my residential network with currently 58 users online, one gaming fanatic has reported a huge improvement in speed. He's not getting killed anywhere near so often these days, apparently..


    AMD_RULES Addicted to LI Member

    Been running v.1.19 stable now for 40 days
  89. i1135t

    i1135t Network Guru Member

    Quick question, is the UP-time only as long as the WAN IP lease time? That makes more sense to me as I notice that my UP-time keeps resetting after 3 days... same amount as my DHCP lease. Uhhh! There goes my theory for testing my UP-Time...
  90. LLigetfa

    LLigetfa LI Guru Member

    No, router uptime is unrelated to WAN uptime. Clear your NVRAM and hand job in all the settings.
  91. i1135t

    i1135t Network Guru Member

    I did just that hoping that would fix it, but nope, still rebooting. I even downgraded from the SpeedMod to the original tomato firmware. I guess I will just wait to see if the 1.20 version corrects the problem.
  92. StevenG

    StevenG LI Guru Member

    I had about 58 days of uptime, but an extended power outage last night killed that record! My UPS shut the PC down and ran out of juice after 15 minutes.

    1.19 is great!
  93. i1135t

    i1135t Network Guru Member

    Ah.... well, I did some tests by using a seperate AP for wireless, while disabling the wireless on the router... and behold!, have 4 days up time, the longest for me yet?!?! Hmmm, maybe a flaw in the wireless portion of the firmware or maybe hardware related? Will keep you guys posted and see what happens...
  94. occamsrazor

    occamsrazor Network Guru Member

    Coming at this late in the game, but I just noticed the technique mentioned previously in this thread:

    #nvram set qos_pfifo=1
    #nvram commit

    Ping time to before making the change was 180ms, afterwards it's 70ms. Seems like a big improvement. This is on the "Tomato Mod v1.19.1464 with OpenVPN" mod.


  95. Toastman

    Toastman Super Moderator Staff Member Member


    I see you're still having the same problems with Intel cards on your system? I just swapped out the firmware on all my WRT54GL AP's to DD-WRT mini generic version v24 sp1. This uses Broadcom wireless driver and has effected a complete cure to both reboots and failure of Intel cards to associate. Now we have to wait until this driver is incorporated into Tomato, work is in progress now.
  96. i1135t

    i1135t Network Guru Member

    Yes, I sorta figured it was a issue with the wireless drivers affecting the UPTIME. It's OK, luckily I had another AP to use, so it'll do for now. Until the wireless drivers are updated on the tomato firmware, I will use it as is for now.

    Good to know it's a work in progress... cheers.
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice