IO stress utility failed with the messages attached below.
This problem occurs on various types of drives on various controllers. I
encountered this problem during heavy IO activities on drives on cpqfc, IDE
(ATA-100), cpqarray, and cciss drivers.
System: DL320(single PIII 1Ghz, 128MB RAM, 256MB swap)
ML530(Dual PIII 1Ghz, 128MB RAM, 256MB swap)
DL380 G2(Single 1.33Ghz, 256RAM, 512MB swap)
filesystems: ext2, ext3
OS: fairfax Beta2
Controller: SA5300, Embedded IDE(ATA100), SA532, SA5i, SA4200
Steps to reproduce:
1. Install OS
2. Create partitions(I tried 2G to 14G) with ext2 or ext3
3. mount partitions
4. Start IO stress utility(anything that creates heavy IO activities)
Error message from /var/log/messages:
Jul 18 14:58:39 localhost kernel: VM: killing process seq_io-linux
Jul 18 14:58:39 localhost last message repeated 2 times
Jul 18 14:59:12 localhost kernel: VM: killing process rand_io-linux
Jul 18 15:01:11 localhost kernel: JBD: out of memory for journal_head
Jul 18 15:01:14 localhost kernel: VM: killing process seq_io-linux
Jul 18 15:01:32 localhost kernel: VM: killing process rand_io-linux
Jul 18 15:02:54 localhost kernel: VM: killing process diff
Jul 18 15:02:54 localhost last message repeated 9 times
we're aware of the "early kills" of the VM and doing our best to fix them as
soon as possible. Thanks for the report.
This defect is considered SHOULD-FIX for Fairfax.
Roswell beta has this fixed. However it's slow in other respects.
Closing this bug but be assured: VM is a very high attentionpoint for us.