BOINC Manager shows 183 completed Rosetta work units in a status of Uploading yet they do not upload.

Rosetta is the only project this system executes.

The Transfers tab shows these 183 completed WUs are in Upload Pending status.

Messages tab show the following log events individually for each of the 183 WUs. This started July 1 and has continued ever since through today, July 7. The oldest completed WUs start to expire early tomorrow morning.

t0 "Started download of (WU name)"
t0+22 sec "Project communication failed: attempting
access to reference site"
t0+22 sec "Temporarily failed upload of (WU name):
HTTP error"
t0+22 sec "Backing off (delta time) on upload of (WU
name)
t0+24 sec "Internet access OK - porject servers may
be temporarily down."

This continues following system reboots.

All other internet access is AOK, no problems.

How might this problem be further investigated, and remedied? Can this be solved without deleting all or most of the completed WUs?

Thanks much


QueueNut

Which OS? Try disabling the firewall temporarily, if any is running. Just to make sure, it's not the firewall causing the problems.

---

Which OS? Try disabling the firewall temporarily, if any is running. Just to make sure, it's not the firewall causing the problems.

The operating system is Windows XP SP3.

After disabling the Windows firewall, commanded an update in BOINC. Messages state:

"update requested by user"
"Sending scheduler request: Requested by user."
"Not reporting or requesting tasks"
"Scheduler request completed"

No change, no completed WUs uploaded (now >200 of them).

The firewall would not have been expected to be causing a problem. Rosetta has been working fine for a long time.

QueueNut

Which OS? Try disabling the firewall temporarily, if any is running. Just to make sure, it's not the firewall causing the problems.

The operating system is Windows XP SP3.

After disabling the Windows firewall, commanded an update in BOINC. Messages state:

"update requested by user"
"Sending scheduler request: Requested by user."
"Not reporting or requesting tasks"
"Scheduler request completed"

No change, no completed WUs uploaded (now >200 of them).

The firewall would not have been expected to be causing a problem. Rosetta has been working fine for a long time.

QueueNut

Might sound archaic, but have you tried rebooting the PC?

---

Which OS? Try disabling the firewall temporarily, if any is running. Just to make sure, it's not the firewall causing the problems.

The operating system is Windows XP SP3.

After disabling the Windows firewall, commanded an update in BOINC. Messages state:

"update requested by user"
"Sending scheduler request: Requested by user."
"Not reporting or requesting tasks"
"Scheduler request completed"

No change, no completed WUs uploaded (now >200 of them).

The firewall would not have been expected to be causing a problem. Rosetta has been working fine for a long time.

QueueNut

Might sound archaic, but have you tried rebooting the PC?

See the first post in the thread,
"This continues following system reboots."

QueueNut

Under Transfers are those 200+ completed WUs - in perpetual Retry in nn:nn:nn with a project backoff time.

I aborted transfer of first one and then a second WU. BOINC reattempts the upload, fails and injects a gradually increasing project backup time delta for each retry and failure.

"Retry..Upload Pending..Uploading.." then backoff to retry again later (which gradually increases).

Finally after aborted 6-8 of these 200+ pending WU uploads, BOINC began to successfully upload one at a time.

Once again, this problem behavior did not change after system reboot.

I did not deinstall/reinstall the BOINC Manager (BTW does that wipe out all completed and "upload pending" WUs?)


QueueNut

I did not deinstall/reinstall the BOINC Manager (BTW does that wipe out all completed and "upload pending" WUs?)

That's a good question. I didn't dare trying to udpade the BOINC-Manager as long as there are tasks.
I would like to know as well, if this is an option somce this would save me the time it takes to let Rosetta run out of work before updating the BOINC-Manager.

Back to your problem: Did you change anything in your router or add any hosts-entries to the Windows hosts file? Any other changes made to the system?

If this happens again, please try to ping and tracert boinc.bakerlab.org
and srv4.bakerlab.org. This should at least indicate wether this is a local problem or a problem with your provider (sometimes DNS-caching gets messed up).

cu

Joe

---

Do you have any anti-virus or anti-spyware software on your computer? Some anti-virus programmes have file filters that act like mini-firewalls.

---

I did not deinstall/reinstall the BOINC Manager (BTW does that wipe out all completed and "upload pending" WUs?)

That's a good question. I didn't dare trying to udpade the BOINC-Manager as long as there are tasks.
I would like to know as well, if this is an option somce this would save me the time it takes to let Rosetta run out of work before updating the BOINC-Manager.

cu Joe

Yes you can upgrade your Boinc version even if you have work pending. I have NOT tried updating though if I have had any pending upload units, I have always uploaded those then upgrade Boinc. The new versions of Boinc will even stop and exit and then restart Boinc, sometimes after a machine restart, if you forget to do it manually.

Under Transfers are those 200+ completed WUs - in perpetual Retry in nn:nn:nn with a project backoff time.

I aborted transfer of first one and then a second WU. BOINC reattempts the upload, fails and injects a gradually increasing project backup time delta for each retry and failure.

"Retry..Upload Pending..Uploading.." then backoff to retry again later (which gradually increases).

Finally after aborted 6-8 of these 200+ pending WU uploads, BOINC began to successfully upload one at a time.

Once again, this problem behavior did not change after system reboot.

I did not deinstall/reinstall the BOINC Manager (BTW does that wipe out all completed and "upload pending" WUs?)

QueueNut

Write down a work unit number and then go to the webpage and check your completed units to see if that number is already uploaded. Sometimes Boinc gets confused and uploads a unit but forgets to delete it from the local pc. You will have to be careful though, lots of workunits have very similar names!

I did not deinstall/reinstall the BOINC Manager (BTW does that wipe out all completed and "upload pending" WUs?)

That's a good question. I didn't dare trying to udpade the BOINC-Manager as long as there are tasks.
I would like to know as well, if this is an option somce this would save me the time it takes to let Rosetta run out of work before updating the BOINC-Manager.

Back to your problem: Did you change anything in your router or add any hosts-entries to the Windows hosts file? Any other changes made to the system?

If this happens again, please try to ping and tracert boinc.bakerlab.org
and srv4.bakerlab.org. This should at least indicate wether this is a local problem or a problem with your provider (sometimes DNS-caching gets messed up).

cu

Joe

Hi Joe -

No changes to hosts file, no changes to router, for >1 year.
Installed Western Digital (WD) disk diagnostics and Acronis True Image WD Edition. Without other evidence, must conclude this software is irrelevant to BOINC and Rosetta.

Both bakerlab.org FQDNs you mention ping successfully, as expected.

Update on behavior last night: After aborting transfer (upload) of a handful of completed WUs, BOINC uploaded many but not all completed WUs. It stopped before the last ~50+ were uploaded.

Update late this afternoon: Completed WUs are piling up again, 100+ now. Retry of transfers does not respond successfully. Messages once again state:

"Started upload of (WU name)"
"Project communication failed: attempting access to reference site"
"Temporarily failed uploaded of (WU name):HTTP error"
"Backing off (time delta) on upload of (WU name)"
"Internet access OK - project servers may be temporarily down."

Pinged those two bakerlab.org FQDNs again, successfully.

Aborted a transfer of a completed WU. Retry Transfer. Failed again, same error messages as above. Tried this same sequence one at a time, 8 WUs, no change.

Also observed no new WUs are being downloaded. Only 5 WUs remain to be processed - yet preferences are set for a queue of 4 days worth (Core i7 920 processor).

Commanded an Update at Project tab. This reported 3 completed WUs in the Messages log. Remaining are 100 completed WUs to upload, in Retry status.


QueueNut