Planner timeout

[sprbck]

Hello,

I installed latest versions of the software. I now have a RHEL machine backing up several linux machines (without any issue) and ONE windows client giving a tremendous headache.

The windows client always times out doing the estimate (etimeout = 20 mins). As I write this post i'm trying a new dump with 1 hour (!) etimeout, let's see what happens.

But my question is.. is this normal behaviour with windows client? I have a Linux machine with twice the data volume and the planner finishes under 5mins. It's hard to believe that it takes so long on Windows, even on a crowded FS. I first tried to use direct CIFS mount, and yes it was slow, understandably, so i ditched it in favor of the Windows client.

I'd like some feedback from you guys, because i'm starting to think i need another solution for backing up windows machines, i can't have a machine thinking for 1 hour just before starting the backup.

Any thoughts appreciated.

Thank you,
Rodrigo

[paddy]

Please use "strategy calcsize" in the dumptype (http://wiki.zmanda.com/index.php/Amanda.conf#DUMPTYPE_SECTION). This will reduce the estimate time for Windows client.

Paddy

[Andrew Rakowski]

Please use "strategy calcsize" in the dumptype (http://wiki.zmanda.com/index.php/Amanda.conf#DUMPTYPE_SECTION). This will reduce the estimate time for Windows client.
Paddy

I believe that should be "estimate calcsize" rather than "strategy calcsize", shouldn't it?

I'm currently fiddling with the "estimate" option, trying to get rid of the occasion amandad processes that are getting left behind (I think when the estimates run too long and the process gets whacked). I was trying "estimate calcsize" vs increasing the estimation timeout with the "etimeout 300" directive (which, oddly, is NOT in the dumptype section).

Cheers,

-Andrew

[paddy]

I believe that should be "estimate calcsize" rather than "strategy calcsize", shouldn't it?

-Andrew

Andrew is correct. It should be "estimate calcsize".

Paddy