Physical disk pool almost full (about 90%)

[Fabien Rouach]

Hi,

one of my physical disk went almost full (90%) (not in the default pool)

I'm accessing through FC and LUN stop responding (offline in vmware)

I'm using v3.2.15

disconnected everything in fc and tried stopping service, nothing seems to happen, but my default pool disk usage shows 100% continuously

What could it be actually doing?

is it safe to force shutdown the machine?

I think i had read about the 10% overhead in the past, is there a way to change that, even temporarly so i can do cleanup in the LUN?

Thanks

[QUADStor Support]

You could try btrace and send it across

yum install -y blktrace
btrace -w 60 /dev/sda > /tmp/btrace.out

Replace /dev/sda with the default pool device path. Then send across
/tmp/btrace.out

The disk usage is probably because the service shutdown tries to sync
the deduplication tables back to disk. If the deduplication tables are
sequential this should not take much time but older installations or
installations with SequentialTables=0 in quadstor.conf can take a lot
more time to sync because while the initial tables maybe sequential on
disk the ones created later will not be and can take many minutes to
shutdown or reload the tables into memory.

An improvement for this behavior has gone recently into the VTL
product and we will be implementing this for the virtualization
product in a week or two and this should solve the slow shutdown/load.
We were able to bring down the load time from some of the extreme
cases (60+ minutes) to under 10 miutes

> --
> You received this message because you are subscribed to the Google Groups "QUADStor Storage Virtualization" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to quadstor-vir...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/quadstor-virt/d2eae24c-ea6a-40f0-b7b6-186c327ad946%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.