Transmission causes leopard slow death

[parodyr]

Now, it's back to the standard setting (No Limit for Upload, 100KiBi for download) and it works like a charm...

That should say:

"No Limit for download, Speed-Limit 100KiBi for upload."

Sorry...

[oraguru]

Hi guys

I posted a couple of weeks ago when I upgraded from Transmission 1.42 to 1.50. After I did that, after maybe 1 day the Mac (iMac 24", 3006 GHz, 4GB RAM) started to get unstable, Transmission showed the turning wheel and at the end only a hard reset helped... I did that more then 10 times... After a while I restored Transmission 1.42 and never ever had that problem again.

Now I thought, maybe Transmission 1.51 is better, so I installed it and it run for 3 days without trouble, but suddenly the problems started again... I'm downloading around 30 files in parallel, all are > 1GB... If I change from 'downloading' view to 'paused' or 'seeding' it starts to slow down, after 3-4 times, it freezes.... the other apps behave really strange, altough I can switch between spaces and windows, I can't click in any apps, it doesn't react.. .

If I presse command + control + ESC I get the little window to kill single programms... it shows that Transmission is not reacting, but no chance to kill it.... Finder starts to freeze... only solution -> Power On Button ...

I tried a couple of times, last night I restored Transmission 1.42 and till today at least, it runs smoothly and changing between the single views is much much faster then with 1.51 or 1.50.

Does anyone has an idea what the problem could be ?

Thanks a lot
G

[oraguru]

Hey !

I completely forgot to post something which I saw on a post before and now 'springs' to my mind... you guys are right, I think the problem starts when I click the 'turtle' at the bottom.. I start the speed limit (download None, Upload 80Kb/s).. then for sure I have a problem...

Somehow I can't get rid of the feeling that 1.42 reacted much faster to user interactions...

But yes with Speedlimit I definately have the freeeze problem...

thanks guys

[kidmidnight]

ive been able to change the speed limit by pausing all torrents, changing the limit (lets say to limit downloads to 100), then resuming all. This seems to do the trick. that said, the problem is so intermittent that this could be a red herring. seems to be working for me right now though.

[transfury]

Running Leopard 10.5.6 I experienced the same type of problems running only 1.42, which had been operating fine for the past few weeks. Today I got the spinning beachball after which Transmission crashed. On relaunch, it eventually caused the computer to freeze repeatedly--the clock on the toolbar froze to confirm that. The problem continued to get worse so that the computer froze on starting up Transmission--it didn't even get a chance to display any panes.

I decided to try the easiest solution and switch to Transmission 1.50 which I already had waiting on my hard drive. On first launch it lasted about a minute before the program crashed. I relaunched, but this time quickly removed one of the items I was downloading, since people appeared to have problems with four simultaneous downloads. Once I reduced the downloads to three, the crashing and freezing stopped.

For what it's worth.

[kidmidnight]

Tested it out by changing the limit while Transmission was running. Slow death started soon after. This problem has affected every transmission release on leopard. I think parodyr has it right.

[oraguru]

Hi all

I can't completely agree with kidmidnight with 'This problem has affected every transmission release on leopard'..

With me it affected all the Transmission releases from 1.50 on, with the ones before (ok I only used 1.42) it never did. With 1.42 (which I went back to) I can start/stop the speed limit, I can change it, I can switch view (all, active, downloading, seeding, paused) no problem, it does it really fast. If I check the CPU consumption it shows always around 15-17% (iMac 3.06 GHz).

I have 25 torrents in parallel right at this moment, 11 downloading and all the others seeding... I do not have one file which is smaller than 1GB, all are at least 1.5 or larger (24GB).

As I wrote before, as soon as I use 1.50 or 1.51 and start switching Speed limit on/off, I get the freeze..

Thanks
G

[kidmidnight]

thats wonderful that it hasnt happened to you, but if you would read the first 19 pages you see the same thing has been happening to people with every release since leopard came out.

[oraguru]

Hi kidmidnight

you're right I'm sorry... I read a bit trough the last 19 pages, but fortunately I can say that with my configuration 1.42 doesn't crash or slows down Leopard... it never did... as soon as I have 1.50 or above, the problems starts...

1.42 is running now again since about 5 days, day and night, downloading and seeding and no problem... I'm working on the same machine every night... no problem..

So if I can be of any assistance in giving you information details about my configuration, maybe someone finds a difference...

I think there are a lot of people out there running 1.42 without a problem, they just don't say anything, as they have no problems... so what probably would help is finding out the difference between our configuration and compare it with the ones who have troubles... maybe we find something like that, together...

Just tell me what would help to know and I'll deliver...

just a thought....

[guilherme]

Both success and failure in reproducing this bug are important. It's not because it didn't happen to him that it does not exist. On the other hand, maybe because it does not happen on his setup, someone might be able to find a fix for everyone's...
I just feel like there's need to comment in a rather provocative way, specially when it's about helping everyone who faces this bug.

[transfury]

Running Leopard 10.5.6 I experienced the same type of problems running only 1.42, which had been operating fine for the past few weeks. Today I got the spinning beachball after which Transmission crashed. On relaunch, it eventually caused the computer to freeze repeatedly--the clock on the toolbar froze to confirm that. The problem continued to get worse so that the computer froze on starting up Transmission--it didn't even get a chance to display any panes.

I decided to try the easiest solution and switch to Transmission 1.50 which I already had waiting on my hard drive. On first launch it lasted about a minute before the program crashed. I relaunched, but this time quickly removed one of the items I was downloading, since people appeared to have problems with four simultaneous downloads. Once I reduced the downloads to three, the crashing and freezing stopped.

OK, discount that. The program was kvetching not because there were four downloads in the queue but because one of the torrents was suffering from "input output error"--a 4 GB download that was causing the program grief. Just happened to be the one I removed.

[vmdeep]

I get such problem, when upgrade from 10.5.5 to 10.5.6. Before this, transmission running without any problem (1.51). I think, may be, problem in Leopard settings?

[Trimalchio]

I just started getting these problems again, after upgrading to 1.5.1, random crashes coming out of nowhere, uTorrent mac didn't do anything of the sort for the few days I was running it. I tried replacing Transmission with 1.4.2 but that kept doing it too. Now I don't know what other steps there are to take. uTorrent sucks compared to transmission but it doesn't destroy my computer, and the weird thing is that I've been running transmission 24/7 for 8ish months with 200 torrents and no problems! It just starts doing it again now.
What are the steps to try and get transmission back to a working state?

[livings124]

Trimalchio: http://forum.transmissionbt.com/viewtop ... f=4&t=3274 specifically #6

[Trimalchio]

So my crash is becoming more and more frequent, but I've noticed a pattern, each fail seems to begin as soon as ctl_enqueuedata: m_allocpacket_internal fails. This time directory service crashed and came back twice and it did it again each time. Also, its funny how this happens, because I see that the sign of failure comes at one time, but at the time I was watching a video off of my external hard drive, which is where torrents go to, and the video didn't so much as stutter, but other operations like brining the dock back up and doing anything in finder crash the respective applications. Is there a way to get extra output from transmission so we can see where it goes wrong more closely? I'd be interested in doing anything I can to help figure out what these crashes are caused by. Is there any other action I can take other than just not using transmission? I really hate that its causing my system to die, because its the best bt client for mac by a long shot.

This is the start of the problem

Mar 8 22:26:01 Rincewind kernel[0]: ctl_enqueuedata: m_allocpacket_internal(8) failed
Mar 8 22:26:01 Rincewind kernel[0]: ALF ALERT: sockwall_cntl_releasecachedpath ctl_enqueuedata rts err 55
Mar 8 22:26:58 Rincewind kernel[0]: ctl_enqueuedata: m_allocpacket_internal(8) failed
Mar 8 22:26:58 Rincewind kernel[0]: ALF ALERT: sockwall_cntl_releasecachedpath ctl_enqueuedata rts err 55
Mar 8 22:30:04 Rincewind DirectoryService[11]: CCachePlugin::WaitForDNSQueries - aborting because cancelled threads didn't return after 30 seconds.
Mar 8 22:30:05: --- last message repeated 3 times ---
Mar 8 22:30:05 Rincewind ReportCrash[13554]: Formulating crash report for process DirectoryService[11]
Mar 8 22:30:05 Rincewind com.apple.launchd[1] (com.apple.DirectoryServices[11]): Exited abnormally: Abort trap
Mar 8 22:30:05 Rincewind DirectoryService[13557]: Launched version 5.6 (v514.24)
Mar 8 22:30:05 Rincewind DirectoryService[13557]: Improper shutdown detected
Mar 8 22:30:06 Rincewind ReportCrash[13554]: Saved crashreport to /Library/Logs/CrashReporter/DirectoryService_2009-03-08-223004_localhost.crash using uid: 0 gid: 0, euid: 0 egid: 0
Mar 8 22:26:01 Rincewind kernel[0]: ctl_enqueuedata: m_allocpacket_internal(8) failed
Mar 8 22:26:01 Rincewind kernel[0]: ALF ALERT: sockwall_cntl_releasecachedpath ctl_enqueuedata rts err 55
Mar 8 22:26:58 Rincewind kernel[0]: ctl_enqueuedata: m_allocpacket_internal(8) failed
Mar 8 22:26:58 Rincewind kernel[0]: ALF ALERT: sockwall_cntl_releasecachedpath ctl_enqueuedata rts err 55
Mar 8 22:30:04 Rincewind DirectoryService[11]: CCachePlugin::WaitForDNSQueries - aborting because cancelled threads didn't return after 30 seconds.
Mar 8 22:30:05: --- last message repeated 3 times ---
Mar 8 22:30:05 Rincewind ReportCrash[13554]: Formulating crash report for process DirectoryService[11]
Mar 8 22:30:05 Rincewind com.apple.launchd[1] (com.apple.DirectoryServices[11]): Exited abnormally: Abort trap
Mar 8 22:30:05 Rincewind DirectoryService[13557]: Launched version 5.6 (v514.24)
Mar 8 22:30:05 Rincewind DirectoryService[13557]: Improper shutdown detected
Mar 8 22:30:06 Rincewind ReportCrash[13554]: Saved crashreport to /Library/Logs/CrashReporter/DirectoryService_2009-03-08-223004_localhost.crash using uid: 0 gid: 0, euid: 0 egid: 0
Mar 8 22:30:43 Rincewind kernel[0]: ctl_enqueuedata: m_allocpacket_internal(8) failed
Mar 8 22:30:43 Rincewind kernel[0]: ALF ALERT: sockwall_cntl_releasecachedpath ctl_enqueuedata rts err 55
Mar 8 22:30:43 Rincewind kernel[0]: ctl_enqueuedata: m_allocpacket_internal(8) failed
Mar 8 22:30:43 Rincewind kernel[0]: ALF ALERT: sockwall_cntl_releasecachedpath ctl_enqueuedata rts err 55
Mar 8 22:32:18 Rincewind kernel[0]: ctl_enqueuedata: m_allocpacket_internal(8) failed
Mar 8 22:32:18 Rincewind kernel[0]: ALF ALERT: sockwall_cntl_releasecachedpath ctl_enqueuedata rts err 55
Mar 8 22:42:10 Rincewind com.apple.launchd[1] (org.samba.smbd[2520]): Stray process with PGID equal to this dead job: PID 13561 PPID 1 smbd
Mar 8 22:42:10 Rincewind com.apple.launchd[1] (org.samba.smbd[2520]): Stray process with PGID equal to this dead job: PID 2521 PPID 1 smbd
Mar 8 23:01:24 localhost kernel[0]: npvhash=4095 //This is when I did a hard reset on my computer.
Mar 8 23:01:19 localhost com.apple.launchctl.System[2]: fsck_hfs: Volume is journaled. No checking performed.
... and so on the boot continues.

Also, I'm using 1.4.2, though this happened on 1.5.1 as well.
I'm on OS X 10.5.6 on a iMac 2.4 with 3 gb ram and an external FW800 drive that the torrents go to.
The torrents I've got downloading include a 200gb torrent that I'm only getting part of, and a ~170 other torrents seeding.
(The external drive recently started getting closer to full, have others noticed that it happens more on close to full drives?)
(Also I'm gonna try the nightly builds and see if the same thing happens)

EDIT:
So with the same torrents running, I used the the nightly build r8040 and it did it again within an hour. This time the above system log entries didn't precede the crash, though I didnt give the crash long before i reset the system, I dont know if the above problems are a symptom of the problem or a cause, but this may indicate that its more of a symptom.