Development for LHC@home

Don't know if this has to do with changes to the Alpha 7.6.15 Release

Don't know if this has to do with changes to the Alpha 7.6.15 Release

I doubt it. That comes from us, not BOINC:
/cvmfs/cms.cern.ch/CMS@Home/agent/CMSJobAgent.sh
echo $(date +"%R:%S %z %Y-%m-%d") "[INFO] Volunteer: ${USERNAME} (${USERID}) Host: ${HOSTID}"
if ! [[ ${USERID} =~ ^[0-9]+$ ]]; then
echo $(date +"%R:%S %z %Y-%m-%d") "[INFO] BOINC_USERID ${USERID} is not an integer"
echo $(date +"%R:%S %z %Y-%m-%d") "[INFO] Going to sleep for 1 hour"
touch /home/boinc/lock
exit 1
fi

Your user-data is absent, which implies to me that the floppy-disk image in your VM may be corrupt. I'd suggest resetting the project to download a fresh VM.

---

I was talking with David and Rom about it, finally Rom thought it must have been your part.

Found this code-snippet:

14:58:01 +0100 2015-11-12 [INFO] Starting CMS Application - Run 7
14:58:01 +0100 2015-11-12 [INFO] Reading the BOINC volunteer's information
14:58:02 +0100 2015-11-12 [INFO] Volunteer: () Host:
14:58:02 +0100 2015-11-12 [INFO] BOINC_USERID is not an integer
14:58:02 +0100 2015-11-12 [INFO] Going to sleep for 1 hour
16:00:01 +0100 2015-11-12 [INFO] Starting CMS Application - Run 8
16:00:01 +0100 2015-11-12 [INFO] Reading the BOINC volunteer's information
16:00:02 +0100 2015-11-12 [INFO] Volunteer: () Host:
16:00:02 +0100 2015-11-12 [INFO] BOINC_USERID is not an integer
16:00:02 +0100 2015-11-12 [INFO] Going to sleep for 1 hour
17:02:50 +0100 2015-11-12 [INFO] Starting CMS Application - Run 1
17:02:50 +0100 2015-11-12 [INFO] Reading the BOINC volunteer's information
17:02:51 +0100 2015-11-12 [INFO] Volunteer: Yeti (250) Host: 495

Since then it is working again

I was talking with David and Rom about it, finally Rom thought it must have been your part.

Found this code-snippet:

16:00:01 +0100 2015-11-12 [INFO] Starting CMS Application - Run 8
16:00:01 +0100 2015-11-12 [INFO] Reading the BOINC volunteer's information
16:00:02 +0100 2015-11-12 [INFO] Volunteer: () Host:
16:00:02 +0100 2015-11-12 [INFO] BOINC_USERID is not an integer
16:00:02 +0100 2015-11-12 [INFO] Going to sleep for 1 hour
17:02:50 +0100 2015-11-12 [INFO] Starting CMS Application - Run 1
17:02:50 +0100 2015-11-12 [INFO] Reading the BOINC volunteer's information
17:02:51 +0100 2015-11-12 [INFO] Volunteer: Yeti (250) Host: 495

Since then it is working again

Yes, looks like something reset and then the user-info became available again. Perhaps a new glide-in starting?

---

Yes, looks like something reset and then the user-info became available again. Perhaps a new glide-in starting?

A reboot of the whole system

Yes, looks like something reset and then the user-info became available again. Perhaps a new glide-in starting?

A reboot of the whole system

Ah!

[Edit]There have been some weird failures since we started up today. E.g. "software version not available", "module bin2ascii not found" (it's part of the standard python library), and the above. Perhaps not having jobs available leads to some sort of corruption in the VM -- although if "a reboot of the whole system" cures it, it would appear to be the in-memory -- or in-slot -- copy rather than the master on-disk instance. [/Edit]

---

I'm not sure how things are handled in the VM these days, so a lot might have changed.

There used to be a bug in the readFloppy.pl script that would attempt to open the floppy device in read/write mode and would fail. This lead to an error on line 55 or 58, I don't remember specifically. It has been a few years since I first looked into it.

For one reason or another when a VM booted under VirtualBox it would flag the floppy device as read-only. This would lead to perl complaining about the device being ready only and not being able to open it for read/write. That in turn lead to the host and user id detection failing.

Does the readFloppy.pl script still attempt to convert XML to JSON in place?

----- Rom