High CPU utilisation/ strange error

[leigh]

We're seeing really high CPU usage on our server when accessing Scalix via either IMAP or SWA. I'm also seeing it whilst using SAC, but not Outlook.

Not sure wether it's related, but I'm also getting this rather strange-sounding error:

Code: Select all

ERROR                   Background Sea(Background Sea) Tue Feb 28 09:46:02 2006
[OM 16418] Failure while filtering with the transaction file ~/data/0000089/0006rs3
Pid of logging process: 413


ERROR                   Background Sea(Background Sea) Tue Feb 28 09:46:02 2006
[OM 28875] Attempt to read a block which does not exist from a blocked item.
Pid of logging process: 413


SERIOUS ERROR           Background Sea(Background Sea) Tue Feb 28 09:46:02 2006
[OM 10270] Process about to terminate due to error.
Signal (Segmentation Violation) trapped by process 413
Procedure trace follows:
  <- tf_GetINT16
  -> tf_ReadRecord
  -> tf_GetINT32
  <- tf_GetINT32
  <- tf_ReadRecord 46010 16
  -> tf_GetINT32
  <- tf_GetINT32
  -> tf_GetINT32
  <- tf_GetINT32
  -> ct_ReadExtension 3002 len=73 rec=322
  <- /build/10.0.0.175/src/lib/ombase/sfl/sfl_Blcked.c
  <- /build/10.0.0.175/src/lib/ombase/sfl/sfl_Blcked.c
  <- /build/10.0.0.175/src/lib/ct/ct_rdext.c
  <- /build/10.0.0.175/src/lib/ct/ct_rdext.c
  <- /build/10.0.0.175/src/lib/flt/flt_apfset.c
  <- /build/10.0.0.175/src/lib/flt/flt_apfltp.c
Pid of logging process: 413


SERIOUS ERROR           Background Sea(Background Sea) Tue Feb 28 09:46:02 2006
[OM 10272] BACKTRACE:
/opt/scalix/lib/libom_er.so(er_add_backtrace+0xb6)[0x94ae76]
/opt/scalix/lib/libom_er.so[0x94b162]
/opt/scalix/lib/libom_er.so(er_DumpProcAndExit+0x1f)[0x94b31f]
[0x751420]
/opt/scalix/lib/libom_er.so(er_LogMsg+0x25)[0x94d9c5]
/opt/scalix/bin/search.process[0x804a033]
/opt/scalix/bin/search.process[0x804a32f]
/opt/scalix/bin/search.process[0x804aef9]
/lib/libc.so.6(__libc_start_main+0xc6)[0x255de6]
/opt/scalix/bin/search.process[0x8049739]
^A
Pid of logging process: 413



So, 1: Any idea what may be sllowing things down?
2: What is a "Background Sea" error?

[tonyn]

Is that a "spike" of high CPU, or is it sustained for a length of time (if so how long) ?
Is the high CPU during logging-in or normal usage ?
Does it affect just one user, or all users ?
What process does "top" show as being the culprit(s)? Perhaps you can post the output?

The error log you posted was for the "Background Search Service" e.g. a client doing a search for a message in their mailbox, might invoke this. It may not be relevant to the high CPU. Do you see lots of these errors in the log or was that a one off?

Cheers
Tony
(Scalix Software Engineer)

[leigh]

Found the reason for the high CPU.
The machine is doing large amounts of swap. We only have 512MB of RAM on this machine, and it looks like it's not really sufficient.
vmstat shows around 1000 or so blocks being swapped out every 5 seconds while it's occuring, so it looks like it's definitely a swap problem. More RAM coming right up.