2.4.2 oddity
Dave Williams
dave at clues.com
Wed Jan 19 17:35:48 UTC 2005
I wonder if anybody has any idea based on the kind of symptoms I'm
seeing at the moment? The box in question is a DL380G4 (duel CPU)
with 4GB of RAM running Solaris 9 x86 and an MSA500 bolted on for
disk.
We're running with the array divided into 2x200GB raw devices.
INN (built with largefile support) currently has three innfeed
processes each servicing between 10 and 15 boxen apiece. Periodically
I'm seeing one particular innfeed process grow beyond two gig in
size before failing with
[ID 801587 news.error] failed to malloc 4226867292 bytes at cnfs/cnfs.c
line 1354: Not enough space
Which seems like a massive amount for the thing to be trying to grab -
We're also seeing occasional instances of :-
[ID 385879 news.error] CNFS-sm: could not read token
@03014E4557532D3100000815D1C900000012@ NEWS-1:0x102ba3920c:18:
Invalid argument
[ID 768526 news.error] Could not retrieve
@03014E4557532D3100000815D1C900000012@: read failed
Which leaves me wondering if the spool has got into a flakey state.
I've been staring at the thing for so long that I'm probably missing
something obvious, but any clue would be gratefully received.
Thanks
Dave
--
Dave Williams
dave at clues.com
More information about the inn-workers
mailing list