 |
 |
Hello,
I discovered the reason of a problem, we had for a little while.
We run machines using xrootd 4.3.0 data servers and a redirector on debian jessie(8.3) for our ALICE SE's (storage elements).
All of our data-servers are using a underlying mounted distributed Lustre, which is why we configured our redirectors/data-manager 's config with:
cms.dfs limit 0 lookup distrib mdhold 0 redirect immed
But unfortunately, the ALICE Monitoring Software "Monalisa" still reported a multiple of our storage available(storage_available * number_of_data-servers), even after changes to prevent that from happening in 4.3.0 (like in XrdCmsCluster::Space for example )
So, I dug a little deeper and found out, that Monalisa calls a local "sensor" ( which we cannot change ...),
which uses "xrd data-manager:1094 queryspace /" periodically to get storage information.
This in turn finally led me to XrdClient/XrdClientAdmin.cc:
// Compute an estimation of the available free space in the given cachefs partition
// The estimation can be fooled if multiple servers mount the same network storage
bool XrdClientAdmin::GetSpaceInfo(const char *logicalname,long long &totspace,long long &totfree,long long &totused,long long &largestchunk)
Which is the root of our problem.
I might try and code around a bit next week.
The best thing would be, if the client could query the redirector's cms daemon, for that "isDfs"-flag and only iterate over all hosts if it is unset.
But I actually don't know if and how that can be done (or what problems might occur, if it would be implemented).
Which is why I think someone more experienced can do this or at least tell me how it is done.
Regards,
JK
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or mute the thread.
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-DEV list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-DEV&A=1