I am finding zfs to be poor at dealing with hardware faults, to the point it completely hangs and no utility can be used until the system is rebooted. Is zfs considered production level?
I am working around connectivity faults between some external SATA drives and a server machine, connected over USB3 or eSATA in multiplex. These issues remain a mystery, but in the face of connectivity issues,
zpool commands block forever in any terminal they are run.
In this instance, I simply attempted to
ls a mounted zfs pool/tank and that terminal hung. A new terminal (
Alt+F2) allowed me to attempt a zpool status, this is also hung. Another new terminal (
Alt+F3) I have run
top and can see
txg_sync with 3% CPU usage, and an endless list of
z_rd_int_x processes, each with 0.3% CPU usage. A fourth terminal (
Alt+F4) is an attempt at
zpool iostat, it is also hung.
Periodic messages of:
[tttt.ttttt] INFO: task bash:xxxx blocked for more than 120 seconds. [tttt.ttttt] INFO: task txg_sync:xxxx blocked for more than 120 seconds. [tttt.ttttt] INFO: task zpool:xxxx blocked for more than 120 seconds.
appear. The machine is still serving files from other pools over SAMBA.
How is a supposedly robust mass storage implementation able to get so crippled? How can I gracefully work the problem without just rebooting?
- OS: CentOS 7
- CPU: Intel Core i7 4770K 1150 RB Quad Core 3.5GHz
- Memory: 32GB non ECC
- Drives: 1TB WD Red WD10EFRX SATA 3.5"
dmesg output is vast, you may need to tell me what you're looking for.
Please ask for more information as required.