Hi List
I'm having a really strange problem with one of my amanda clients
(that also happens to be the server).
It's got 2 disklists (separate filesystems), / and /usr
amcheck reports all ok
estimates come back fine.
/ backs up fine
/usr aborts as soon as it comes time to dump the filesystem.
This is with both versions of amanda 2.5.0.p2 and 2.5.1.p2 (a recent
upgrade).
I'm using Kerberos authentication (thanks Amanda team for supporting
this it's great!)
So I get this
qtvsrv1.xxxx:/ 0 221m finished (22:22:38)
qtvsrv1.xxxx:/usr 0 driver: (aborted:"[request failed:
error sending REQ: krb5 write error to qtvsrv1.xxxx: Broken pipe]")
(too many dumper retry)
I can't find any relevant debug info at all that even suggests it's
attempted to connect, which is really strange. From the 'log' file :
FAIL chunker qtvsrv1.xxxx /usr 20070202 0 [cannot read header: got 0
instead of 32768]
CHUNK taper qtvsrv1.xxxx / 20070202 1 0 [sec 6.862 kb 227136 kps
33098.2 {wr: writers 7098 rdwait 0.000 wrwait 4.706 filemark 2.081}]
CHUNKSUCCESS taper qtvsrv1.xxxx / 20070202 0 [sec 6.862 kb 227136 kps
33098.2 {wr: writers 7098 rdwait 0.000 wrwait 4.706 filemark 2.081}]
FAIL driver qtvsrv1.xxxx /usr 20070202 0 [too many dumper retry:
"[request failed: error sending REQ: krb5 write error to
qtvsrv1.xxxx: Broken pipe]"]
FAIL chunker qtvsrv1.xxxx /usr 20070202 0 [cannot read header: got 0
instead of 32768]
Has any one any ideas ?? This is the only host with a problem, out of
20 or so.
It's driving me mad !
---
AlanP
