--
Bernd Lentes
System Administrator
Institute for Metabolism and Cell Death (MCD)
Building 25 - office 122
HelmholtzZentrum München
bernd.lentes@helmholtz-muenchen.de
phone: +49 89 3187 1241
phone: +49 89 3187 3827
fax: +49 89 3187 2294
http://www.helmholtz-muenchen.de/mcd
----- On Feb 25, 2021, at 5:43 PM, Wengang Wang wen.gang.wang@oracle.com wrote:
On Feb 25, 2021, at 4:39 AM, Lentes, Bernd < [
mailto:bernd.lentes@helmholtz-muenchen.de | bernd.lentes@helmholtz-muenchen.de
] > wrote:
----- On Feb 24, 2021, at 10:49 PM, Wengang Wang [
mailto:wen.gang.wang@oracle.com | wen.gang.wang@oracle.com ] wrote:
On Feb 24, 2021, at 1:11 AM, Lentes, Bernd < [
mailto:bernd.lentes@helmholtz-muenchen.de | bernd.lentes@helmholtz-muenchen.de
] >
wrote:
----- On Feb 23, 2021, at 11:24 PM, Wengang Wang [
mailto:wen.gang.wang@oracle.com |
wen.gang.wang@oracle.com ] wrote:
#define EEXIST 17 /* File exists */
You may already have the file created from the other node.
wengang
Hi Wengang,
thanks for your help.
The reflink is created via:
reflink -fv /path/to/domain.raw /mnt/ocfs2/reflink
Of course this file exists already because the skript ran also one day before.
But from which node the previous one was created is unpredictable, it might be
the same
but also the other one.
Is the message severe or don't i have to bother about it ?
OCFS2 is a shared-storage cluster FS. For the reflinks, you need and should
create them only from one of the nodes. Don’t do duplicated work (reflink) on
other nodes.
Hi Wengang,
thanks for your help.
Just to be sure that i understand you completely:
we reflink some raw file from host 1 and OTHER raw files (not the same !) from
host 2.
Is that already a problem ? Or are you talking about reflinking the SAME file
concurrently from both nodes ?
No, that’s not a problem. Make sure your applications does the same as you want.
From the error message, your app could be creating duplicated reflinks from one
or two nodes.
That’s what we can know by the error message. I am suspecting the problem is in
your app, you could try to debug it.
thanks,
wengang
Hi,
what we do is to overwrite the reflink from the day before with the current one, with the same name.
For this we use reflink -f. Could that be the culprit ?
Should we delete the old one before ? Or maybe use unique names for each reflink ?