Hi,
we observe a problem exporting a lustre directory over nfs version 3. MDS and OSTs are running the lustre kernel: 2.6.16-27-0.9_lustre-1.4.11smp on SLES 10 systems. All lustre clients are using:
2.6.18-8.1.8.el5_lustre.1.6.2smp.
From one of these clients we try to export a directory and have the following entry in /etc/exports:
/lustre/home/zrsdw01 ip_address(rw,no_root_squash,async)
The mounting hosts are running RHEL4 or SuSE 10.1. We have the stanza:
name of exporting host:/lustre/home/zrsdw01  /zrsdw01 nfs bg,intr      0 0
The behavior is unchanged if we add the async option or not. We are using 64bit Systems only.
Exporting, mounting and reading from the freshly mounted dir work fine.
Writing to existing files seems to work without any errors. But closing a vi session resulted in the message:
fsync.tst" E667: Fsync failed
We wrote a small program to get more insight to this problem. To summarize the result: Overwriting an existing file works. Using open(...,O_CREAT,...) works, writing 4Bytes succeeds, close and fsync fail. If we use the sequence fsync, close we get a permission denied from fsync but not from close, if we use close only we get the permission denied from the call to close. (The reported errno was 13) Do you have any idea what is going wrong. What additional information do you need to send us a hint how we can resolve the problem. Are there incompatibilities between the different software components?
thanks
w.dilling

--------------------------------------------------------------------------------
W.Dilling                               Tel.: (49) 7071/29-70206
Universitaet Tuebingen                  Fax.: (49) 7071/29-5912
Zentrum fuer Datenverarbeitung          mail: [EMAIL PROTECTED]
Waechterstrasse 76
72074 Tuebingen

Attachment: smime.p7s
Description: S/MIME krytographische Unterschrift

_______________________________________________
Lustre-discuss mailing list
[email protected]
https://mail.clusterfs.com/mailman/listinfo/lustre-discuss

Reply via email to