[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] problem accessing files on samba

On 5/23/07, Todd Tannenbaum <tannenba@xxxxxxxxxxx> wrote:
Matei Stroila wrote:
> Dear all,
> I am using Condor 6.8.4 and a pool of Windows 2003 servers. The
> executable that I submit needs to load files from a Samba network
> drive. I followed all the instructions for network file access (
> run_as_owner, cred server etc). Everything seemed to run fine until I
> looked at the log files and saw that sometimes the executable cannot
> load remote files. This seems to happen randomly. The executable
> processes a bunch of remote files normally and then it cannot load
> another bunch and so on. Moreover, sometimes, I can no longer login to
> some of the servers after submitting the jobs, for a couple of minutes
> or so. Could this be a Condor configuration or a Windows security
> issue (blocking my account after a threshold number of logins in a
> period of time)?

Or, another completely wild guess, is there some limit on the number of
times a user can be logged in on machine X?  Do your machines advertise
multiple VMs (cpu slots)?

Yes, they do.

If so, perhaps your job is running fine on
vm1@foo, then another job starts on vm2@foo logging in as the same user,
and subsequently vm1@foo stops working correctly?

But they run fine, just that they cannot read files from a network
drive. So my own executable log looks like:

" Processing file //mynetworkdrive/filename1 ...
 Processing file //mynetworkdrive/filename2 ...
 Cannot read file //mynetworkdrive/filename3 ...
 Cannot read file //mynetworkdrive/filename4 ...
 Processing file //mynetworkdrive/filename5 ...

Again, if these files were local, they would be all processed. This
"Cannot read file ..." seems to be random.