Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
What causes "g_eli_read_done() failed" errors?
58 views
Skip to first unread message
Matthew X. Economou
Jan 27, 2012, 11:47:43 AM1/27/12
to
I have a server with root on a zpool and with the zpool on a RAID-Z1 made from a set of geli-encrypted providers. Periodically, I get errors like the following:
GEOM_ELI: g_eli_read_done() failed ada3p4.eli[READ(offset=981766144, length=28672)]
It's always the same geom (ada3p4.eli). The provider (ada3p4) never reports any other errors - no read or write failures, no DMA errors, etc. Sometimes these errors occur at boot time, when FreeBSD mounts the zpool, but I also see them when I scrub the zpool.
The server is an old Dell PowerEdge 2400 running FreeBSD/i386 9.0-RELEASE. One complication is that the server only has 1 GiB RAM installed. I've had to tweak the kernel memory limits in order to stabilize the systems. I'm currently running with the following settings:
vfs.zfs.arc_max="8M"
vfs.zfs.vdev.cache="4M"
vm.kmem_size="256M"
vm.kmem_size_max="512M"
The system is otherwise stable even under load.
Would anyone care to speculate on the causes of these errors? How likely is the cause related to kernel memory utilization versus hard disk drive errors?
Best wishes,
Matthew
GEOM_ELI: g_eli_read_done() failed ada3p4.eli[READ(offset=981766144, length=28672)]
It's always the same geom (ada3p4.eli). The provider (ada3p4) never reports any other errors - no read or write failures, no DMA errors, etc. Sometimes these errors occur at boot time, when FreeBSD mounts the zpool, but I also see them when I scrub the zpool.
The server is an old Dell PowerEdge 2400 running FreeBSD/i386 9.0-RELEASE. One complication is that the server only has 1 GiB RAM installed. I've had to tweak the kernel memory limits in order to stabilize the systems. I'm currently running with the following settings:
vfs.zfs.arc_max="8M"
vfs.zfs.vdev.cache="4M"
vm.kmem_size="256M"
vm.kmem_size_max="512M"
The system is otherwise stable even under load.
Would anyone care to speculate on the causes of these errors? How likely is the cause related to kernel memory utilization versus hard disk drive errors?
Best wishes,
Matthew
0 new messages