Roche FLX Software Problem

[Mike Mitchell]

Folks-

Since the December upgrade to the FLX on-instrument and off-instrument software, we cannot make the off-instrument software run properly.

In particular, the FLX software cannot find a utility named "gstat." I found a geophysical statistical program called gstat and installed it, but it is an interactive program -- not what was needed.

Does anybody know what the gstat package is, and where I might get it? We never have needed it in the past. I don't know why we need it now.

Thanks,

Mike Mitchell

[sklages]

Hi Mike,

Folks-

Since the December upgrade to the FLX on-instrument and off-instrument software, we cannot make the off-instrument software run properly.

In particular, the FLX software cannot find a utility named "gstat." I found a geophysical statistical program called gstat and installed it, but it is an interactive program -- not what was needed.

Does anybody know what the gstat package is, and where I might get it? We never have needed it in the past. I don't know why we need it now.

http://linux.die.net/man/1/gstat

You are running a cluster?

cheers,
Sven

[Mike Mitchell]

More On FLX Problem

Hi Mike,

http://linux.die.net/man/1/gstat

You are running a cluster?

cheers,
Sven

Folks-

We are running both a cluster and un-clustered dual quad core Xeon computers.

Thanks,

Mike

[sklages]

Folks-

We are running both a cluster and un-clustered dual quad core Xeon computers.

Thanks,

Mike

Hmm, seems to be "part" of the support tool ...

It has been already there in former versions of the Roche software, but the roche "script maintainers" have decided to expect 'gstat' now in '/opt/bin', for god knows whatever reasons ...
So maybe changing back '/opt/bin/gstat' to a simple 'gstat' in the corresponding script (gsTiCluster.sh?) or cfg file may help if 'gstat' is installed (but it probably is).

cheers,
Sven

[Mike Mitchell]

More Information

Folks-

This is the first time we have run any FLX processing since the December
update of the FLX software.

I cannot find that we have ever has a program named gstat on any of the machines we use for FLX processing.

We are ONLY running FLX in a non-cluster environment on dual-quad-core Dell servers running RHEL. We set the MULTI option to run the analysis on all 8 cores of the server. Maybe the newer software expects to be running on a cluster? that is a Roche-supplied cluster? I don't know.

The actual data set we are working with is from last September. The user wants an "amplicons" work up. So I have done a clean install of version 2.0.01.12p1 of the FLX software on one of the old, now unused, IPARs.

We are trying that out right now.

Has anyone else tried NOT running the new FLX software on a cluster?

Thanks,

Mike Mitchell

[kmcarr]

Has anyone else tried NOT running the new FLX software on a cluster?

We have been running the new (2.3) software for a while now on our cluster without a problem. gstat is installed in /usr/local/bin and has been since the cluster was configured years ago (as noted above it is part of the Ganglia Cluster Monitoring tool). I have also run the gsSupport tool under the new software without problems.

[sklages]

Folks-
We are trying that out right now.

Has anyone else tried NOT running the new FLX software on a cluster?

Thanks,

Mike Mitchell

If you are not running on a cluster, you should check if GS_LAUNCH_MODE is *really* correctly set:

GS_LAUNCH_MODE=MULTI

We do run the software on 32/48 core servers, no cluster environment, and I could not find any problem as you described (and we definitely have not installed gstat :-))

If GS_LAUNCH_MODE is correctly set we should analyze what you are running, AFAIK gstat is *only* invoked when run in cluster environment.


cheers,
Sven