Avoiding jargon in BOINC message

[mo.v]

I see that a BOINC message says

[error] garbage_collect(); still have active task for acked result

I had never met the verb 'to ack' or the adjective 'acked' before and at first thought the meaning was related to 'Ack!' meaning 'Yuk!' ie this task had produced a rubbish or dreadful result. The presence of 'garbage' in the same message seemed to confirm this interpretation. But when I looked the word up I found that it's an abbreviation for 'acknowledged'.

Non-native English speakers are also receiving this message. They already have a hard time because English has a much larger vocabulary than any other language.

Could BOINC please use the full word in this message to give more crunchers a chance to understand it?

[Nicolas]

FWIW, that's an internal error message. If it's appearing often at all, there is a bug. It's an internal situation that should never happen.

An ACTIVE_TASK represents a running workunit. The garbage_collect() function deletes files that aren't used by any workunit anymore, old application versions, etc. What it found here is an active_task for a workunit that was already reported to the server, which is quite an impossible state.

[Nicolas]

Other "interesting" internal errors I found in the code, for your enjoyment (many of which are very unlikely to get):

• State file error: result <name> is in wrong state
• State file error: two tasks in slot <number>
• Can't handle task <name> in scheduler reply
• Got ack for task <name>, but can't find it
• Server requested conditional abort of unknown task <name>
• OS random string generation failed, exiting
• Couldn't add curlEasy handle to curlMulti
• [http_debug] can't rewind post output file
• Persistent file transfer object not found
• GUI RPC bind failed: <error code> _{[this one is common on Linux, if you start client twice]}

[mo.v]

I like the idea of a 'conditional abort' which is perhaps more hopeful than an unrecoverable one!

We realise it's an internal error. It's happening in a new beta application with some very innovative features but still at an early stage of development.

But the message itself comes from BOINC and has been seen on lots of projects. My only concern in this ticket is with the wording. I'd also request that the same change is made to the Got ack for task <name>, but can't find it message that Nicolas found.

[Nicolas]

Replying to mo.v:

I like the idea of a 'conditional abort' which is perhaps more hopeful than an unrecoverable one!

A conditional abort is a way for the project server to ask the client to abort tasks if they haven't started yet.

[Nicolas]

I don't know much about the implications of having an active task for an acknowledged task; but I think this error could be hidden behind the [state_debug] logging flag along with other messages.

[davea]

These message are intended to provide info for developers. Could the original poster please send me the message log that contained this error (and if it's reproducible, turn on the <task_debug> flag and send the resulting log)

[mo.v]

I wasn't running the task that received this message but I've asked the person whose task it was to provide the message log.

[mo.v]

Here are two HADSM3P v.6.01 tasks with this problem on CPDN Beta:

​http://cpdnbeta.oerc.ox.ac.uk/result.php?resultid=23368 ​http://cpdnbeta.oerc.ox.ac.uk/result.php?resultid=23341

The message can't be reproduced because both tasks have already crashed. But they were quoted almost in full on the forum here: ​http://cpdnbeta.oerc.ox.ac.uk/forum_thread.php?id=128

These message logs from Ananas may also be helpful:

​http://oct31.de/message_log.txt

[Nicolas]

Uploading message log here so it doesn't get lost.

[Nicolas]

r16128 seems to have fixed the bug causing this error message.