Hi,
I am using Amanda 2.6.1p2.
I'm currently using Equallogic iSCSI storage for Amanda's holding disk.
My current Amanda server has iptables disabled because this somehow affects
iSCSI multipathing, i.e. if iptables is enabled, only one path works. I have
yet to determine how to get iptables+iscsi multipathing to work. But this is
not why I'm writing.
I have 2 iSCSI volumes setup for Amanda's usage:
[r...@cronos ~]# df -h /data3 /data4
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/cronos-amanda-holdingdisk
886G 116G 726G 14% /data3
/dev/mapper/cronos-amanda-diskbuffer
394G 199M 374G 1% /data4
One volume contains the holding disk, the other the diskbuffer.
I've noticed by running 'iostat' that Amanda gets terribly slow at reading the
data from the holding disk, i.e. the reading speed would be a crawling
0.5Mb/sec. This happens after the backup has been running for a while, e.g. a
day.
The manual solution I came up with is to change the MTU for the network
interfaces:
ifconfig eth3 mtu 9000
ifconfig eth4 mtu 9000
It takes about 30 seconds for the iSCSI paths to fail and become active again.
Once this happens, Amanda starts reading from the holding disk at normal speed
of ~50Mb/sec:
Device: tps MB_read/s MB_wrtn/s MB_read MB_wrtn
dm-7 791.00 51.80 0.00 51 0
When the problem re-occurs, I change the MTU to a different value: if it's
9000bytes, I change it to 1500bytes, and vice-versa.
Does anyone know why this happens? Why does Amanda become slow at reading from
the holding disk residing on an iSCSI volume?
Best,
--
Valeriu Mutu
