Skip to contentSkip to conversation

deadlock with 2.6.9

DISCUSSION Chuck Ebbert 3 participants 5 messages Nov 7, 2004

lore
Share & tools
lore
#0 lore raw
Quoted 2 lines · Chris Stromsoe

I had a third lockup, this time not related to burning a dvd. As before, the bulk of the processes that were hung were cron

Why so many cron processes? Is this normal on your system, or does it
look like cron keeps spawning processes because it gets no response on the sockets?

Quoted 1 line

The box is P3 SMP

Can you try a uniprocessor kernel?

Quoted 1 line

syslog logs to a stripe of two mirrors, built with mdadm.

Get a real RAID controller (3Ware, not some crappy pseudo-RAID junk.) They are
much more reliable than software RAID.

--Chuck Ebbert 07-Nov-04 00:28:44

#1 lore raw

OT: cron filling process table (was: deadlock with 2.6.9)

Hello,

just an thought on cron, not very kernel related:

Quoted 3 lines · In article <[email protected]> you wrote:

Why so many cron processes? Is this normal on your system, or does it look like cron keeps spawning processes because it gets no response on the sockets?

if you have a cron job which is executed very often cron will spawn a new child everytime the deadline is reached. if the client is stuck for some reason (ie. uninterruptiple sleep while accessing a broken ressource) it will soon fill up your systems.

Personally thats why I prefer a cron-like system, which is configured with maximum concurrency (or always serialize the jobs for a given type). This has problems with handing jobs, but it is generally more stable for the system.

Greetings
Bernd

#2 lore raw
In reply to Chuck Ebbert
Quoted 5 lines · Chuck Ebbert
Quoted 1 line · Chris Stromsoe

I had a third lockup, this time not related to burning a dvd. As before, the bulk of the processes that were hung were cron

Why so many cron processes? Is this normal on your system, or does it look like cron keeps spawning processes because it gets no response on the sockets?

I'm guessing so many processes because every time one gets started, it ends up getting stuck in schedule_timeout(), until the system stops spawning new processes. There are many of them because the system was running for a day or so before it became unresponsive.

Quoted 3 lines
Quoted 1 line

The box is P3 SMP

Can you try a uniprocessor kernel?

Boot with nosmp or boot a kernel compiled for up?

Quoted 3 lines
Quoted 1 line

syslog logs to a stripe of two mirrors, built with mdadm.

Get a real RAID controller (3Ware, not some crappy pseudo-RAID junk.) They are much more reliable than software RAID.

I've had more problems with reliable hardware raid controllers than I have with software raid. Your mileage may vary.

-Chris

#3 lore raw
Quoted 9 lines · Chris Stromsoe
Quoted 5 lines · Chuck Ebbert
Quoted 1 line · Chris Stromsoe

I had a third lockup, this time not related to burning a dvd. As before, the bulk of the processes that were hung were cron

Why so many cron processes? Is this normal on your system, or does it look like cron keeps spawning processes because it gets no response on the sockets?

I'm guessing so many processes because every time one gets started, it ends up getting stuck in schedule_timeout(), until the system stops spawning new processes. There are many of them because the system was running for a day or so before it became unresponsive.

That wasn't very clear. The high number of cron jobs is extremely abnormal. There are generally never more than a handful running at any one time, and very rarely more than only the parent cron process. The most frequent cronjob on the system runs every 5 minutes.

-Chris

#4 lore raw
In reply to Chuck Ebbert
Quoted 2 lines · In article <[email protected]> you wrote:

Get a real RAID controller (3Ware, not some crappy pseudo-RAID junk.) They are much more reliable than software RAID.

On what sample do you base this claim?

Generally hardware raid sooner or later makes problems (especially if you run raid5 in degenerate mode or try to rebuild by disk replacing with differen/old signature). Also bus hangs are commonly not very well received by hw raid firmware or drivers.

Issues with software raid are usually quicker to spot.

Greetings
Bernd