Re: kjournald hang on ext3 to ext3 copy

Re: kjournald hang on ext3 to ext3 copy

[Andreas Dilger]

On Jun 16, 2007  08:17 -0500, John Marconi wrote:
> I am running into a situation in which one of my ext3 filesystems is 
> getting hung during normal usage.  There are three ext3 filesystems on a 
> CompactFLASH.  One is mounted as / and one as /tmp.  In my test, I am 
> copying a 100 MB file from /root to /tmp repeatedly.  While doing this 
> test, I eventually see the copying stop, and any attempts to access /tmp 
> fail - if I even do ls /tmp the command will hang.
> 
> I suspect kjournald because of the following ps output:
> PID      PPID   WCHAN:20      PCPU  %MEM  PSR  COMM
> 8847    99 start_this_handle        1.1  0.0  28     pdflush
> 8853    99 schedule_timeout       0.2  0.0   7     pdflush
>  188     1 kswapd                       0.0  0.0  19   kswapd0
> 8051     1 mtd_blktrans_thread   0.0  0.0  22   mtdblockd
> 8243     1 kjournald                    0.0  0.0   0   kjournald
> 8305     1 schedule_timeout        0.0  0.0   2   udevd
> 8378     1 kjournald                    0.0  0.0   0   kjournald
> 8379     1 journal_commit_trans 16.6  0.0   0   kjournald
> 8437     1 schedule_timeout       0.0  0.0   0   evlogd
> 8527     1 syslog                        0.0  0.0   1   klogd
> 8534     1 schedule_timeout       0.0  0.0   0   portmap
> 8569     1 schedule_timeout       0.0  0.0   0   rngd
> 8639     1 schedule_timeout       0.1  0.0  24   sshd
> 8741  8639 schedule_timeout    0.0  0.0   0     sshd
> 8743  8741 wait                        0.0  0.0   9       bash
> 8857  8743 schedule_timeout    4.9  0.0   7         cp
> 8664     1 schedule_timeout       0.0  0.0   0   xinetd
> 8679     1 schedule_timeout       0.0  0.0   0   evlnotifyd
> 8689     1 schedule_timeout       0.0  0.0   0   evlactiond
> 8704     1 wait                           0.0  0.0   1   bash
> 8882  8704 -                            0.0  0.0   2     ps
> 
> If I run ps repeatedly, I always see process 8379 in 
> journal_commit_transaction, and it is always taking between 12% and 20% 
> of processor 0 up.  This process never completes.  I also see process 
> 8847 in start_this_handle forever as well - so I believe they are related. 
> 
> This system is using a 2.6.14 kernel.

Please try to reproduce with a newer kernel, as this kind of problem
might have been fixed already.


Two tips for debugging this kind of issue:
- you need to have detailed stack traces (e.g. sysrq-t) of all the
  interesting processes

- if a process is stuck inside a large function (e.g. 8379 in example)
  you need to provide the exact line number.  this can be found by compiling
  the kernel with CONFIG_DEBUG_INFO (-g flag to gcc) and then doing
  "gdb vmlinux" and "p *(journal_commit_transaction+{offset})", where the
  byte offset is printed in the sysrq-t output, and then include the code
  surrounding that line from the source file

- a process stuck in "start_this_handle()" is often just an innocent
  bystander.  It is waiting for the currently committing transaction to
  complete before it can start a new filesystem-modifying operation (handle).
  That said, the journal handle acts like a lock and has been the cause of
  many deadlock problems (e.g. process 1 holds lock, waits for handle;
  process 2 holds transaction open waiting for lock).  pdflush might be one
  of the "process 1" kind of tasks, and some other process is holding the
  transaction open preventing it from completing.

Cheers, Andreas
--
Andreas Dilger
Principal Software Engineer
Cluster File Systems, Inc.

Re: kjournald hang on ext3 to ext3 copy

[John Marconi]

Andreas Dilger wrote:
> On Jun 16, 2007  08:17 -0500, John Marconi wrote:
>   
>> I am running into a situation in which one of my ext3 filesystems is 
>> getting hung during normal usage.  There are three ext3 filesystems on a 
>> CompactFLASH.  One is mounted as / and one as /tmp.  In my test, I am 
>> copying a 100 MB file from /root to /tmp repeatedly.  While doing this 
>> test, I eventually see the copying stop, and any attempts to access /tmp 
>> fail - if I even do ls /tmp the command will hang.
>>
>> I suspect kjournald because of the following ps output:
>> PID      PPID   WCHAN:20      PCPU  %MEM  PSR  COMM
>> 8847    99 start_this_handle        1.1  0.0  28     pdflush
>> 8853    99 schedule_timeout       0.2  0.0   7     pdflush
>>  188     1 kswapd                       0.0  0.0  19   kswapd0
>> 8051     1 mtd_blktrans_thread   0.0  0.0  22   mtdblockd
>> 8243     1 kjournald                    0.0  0.0   0   kjournald
>> 8305     1 schedule_timeout        0.0  0.0   2   udevd
>> 8378     1 kjournald                    0.0  0.0   0   kjournald
>> 8379     1 journal_commit_trans 16.6  0.0   0   kjournald
>> 8437     1 schedule_timeout       0.0  0.0   0   evlogd
>> 8527     1 syslog                        0.0  0.0   1   klogd
>> 8534     1 schedule_timeout       0.0  0.0   0   portmap
>> 8569     1 schedule_timeout       0.0  0.0   0   rngd
>> 8639     1 schedule_timeout       0.1  0.0  24   sshd
>> 8741  8639 schedule_timeout    0.0  0.0   0     sshd
>> 8743  8741 wait                        0.0  0.0   9       bash
>> 8857  8743 schedule_timeout    4.9  0.0   7         cp
>> 8664     1 schedule_timeout       0.0  0.0   0   xinetd
>> 8679     1 schedule_timeout       0.0  0.0   0   evlnotifyd
>> 8689     1 schedule_timeout       0.0  0.0   0   evlactiond
>> 8704     1 wait                           0.0  0.0   1   bash
>> 8882  8704 -                            0.0  0.0   2     ps
>>
>> If I run ps repeatedly, I always see process 8379 in 
>> journal_commit_transaction, and it is always taking between 12% and 20% 
>> of processor 0 up.  This process never completes.  I also see process 
>> 8847 in start_this_handle forever as well - so I believe they are related. 
>>
>> This system is using a 2.6.14 kernel.
>>     
>
> Please try to reproduce with a newer kernel, as this kind of problem
> might have been fixed already.
>
>
> Two tips for debugging this kind of issue:
> - you need to have detailed stack traces (e.g. sysrq-t) of all the
>   interesting processes
>
> - if a process is stuck inside a large function (e.g. 8379 in example)
>   you need to provide the exact line number.  this can be found by compiling
>   the kernel with CONFIG_DEBUG_INFO (-g flag to gcc) and then doing
>   "gdb vmlinux" and "p *(journal_commit_transaction+{offset})", where the
>   byte offset is printed in the sysrq-t output, and then include the code
>   surrounding that line from the source file
>
> - a process stuck in "start_this_handle()" is often just an innocent
>   bystander.  It is waiting for the currently committing transaction to
>   complete before it can start a new filesystem-modifying operation (handle).
>   That said, the journal handle acts like a lock and has been the cause of
>   many deadlock problems (e.g. process 1 holds lock, waits for handle;
>   process 2 holds transaction open waiting for lock).  pdflush might be one
>   of the "process 1" kind of tasks, and some other process is holding the
>   transaction open preventing it from completing.
>
> Cheers, Andreas
> --
> Andreas Dilger
> Principal Software Engineer
> Cluster File Systems, Inc.
>
>
>   
Andreas,

Thanks for the information.

I am not able to update the entire kernel to a new version for a variety 
of reasons, however I can update certain parts in my system (such as the 
filesystem).  I did a diff of the 2.6.16 kernel against my kernel, and 
the changes to jbd were minimal.  I plan on looking at the latest 
versions of the kernel to determine if anything has changed since 2.6.16.

I took a look at the place that kjournald was stuck - it is in the 
journal_commit_transaction "while (comiit_transaction->t_updates)" loop 
and it is trying to "spin_lock(&journal->j_state_lock).  When I look at 
pdflush, it is also trying to take the journal->j_state_lock.  Do you 
have any tips on finding out which process might own journal->j_state_lock?

Thanks again,
John

Re: kjournald hang on ext3 to ext3 copy

[Andreas Dilger]

On Jun 18, 2007  22:53 -0500, John Marconi wrote:
> Andreas Dilger wrote:
> >Two tips for debugging this kind of issue:
> >- you need to have detailed stack traces (e.g. sysrq-t) of all the
> >  interesting processes
> >
> >- if a process is stuck inside a large function (e.g. 8379 in example)
> >  you need to provide the exact line number.  this can be found by 
> >  compiling
> >  the kernel with CONFIG_DEBUG_INFO (-g flag to gcc) and then doing
> >  "gdb vmlinux" and "p *(journal_commit_transaction+{offset})", where the
> >  byte offset is printed in the sysrq-t output, and then include the code
> >  surrounding that line from the source file
> >
> >- a process stuck in "start_this_handle()" is often just an innocent
> >  bystander.  It is waiting for the currently committing transaction to
> >  complete before it can start a new filesystem-modifying operation 
> >  (handle).
> >  That said, the journal handle acts like a lock and has been the cause of
> >  many deadlock problems (e.g. process 1 holds lock, waits for handle;
> >  process 2 holds transaction open waiting for lock).  pdflush might be one
> >  of the "process 1" kind of tasks, and some other process is holding the
> >  transaction open preventing it from completing.
> 
> I am not able to update the entire kernel to a new version for a variety 
> of reasons, however I can update certain parts in my system (such as the 
> filesystem).  I did a diff of the 2.6.16 kernel against my kernel, and 
> the changes to jbd were minimal.  I plan on looking at the latest 
> versions of the kernel to determine if anything has changed since 2.6.16.

The problem may also be in the ext3 layer and not jbd.

> I took a look at the place that kjournald was stuck - it is in the 
> journal_commit_transaction "while (comiit_transaction->t_updates)" loop 
> and it is trying to "spin_lock(&journal->j_state_lock).  When I look at 
> pdflush, it is also trying to take the journal->j_state_lock.  Do you 
> have any tips on finding out which process might own journal->j_state_lock?

You can enable CONFIG_DEBUG_SPINLOCK in newer kernels and it appears the
spinlock will set the "owner" field to the task struct.  You still need
to get access to this via e.g. "crash" or lkcd or something.

Hmm, it seems this is only set for ppc and s390???  That is how I would
debug this in any case.  The other way (I've done this too many times
in the past) is to look through all of the stack traces and figure out
which ones are in a filesystem context, then check if any of them are
blocked on locks while holding transactions open.  Needs a detailed
understanding of kernel callpaths.

Cheers, Andreas
--
Andreas Dilger
Principal Software Engineer
Cluster File Systems, Inc.