climateprediction.net - 2005-03-17 15:20:02 - SCHEDULER_REPLY::parse(): bad first tag Content-type: text/plain

climateprediction.net - 2005-03-17 15:20:02 - Can't parse scheduler reply

climateprediction.net - 2005-03-17 15:20:02 - Deferring communication with project for 16 minutes and 49 seconds

I got this message in my hyperthread computer. Is there some conflict when I run Boinc with two "processors". I have 1024 MB memory and I don't overclock anything in my computer. I have tested my hard disk and it is works fine. I leave applications in memory while preempted. I just run climateprediction project and nothing else in my boinc program. I don't use other distributed computing programs. I have always used boinc 4.19 in my hyperthread computer. I use PC-cillin 2002 firewall and windows service pack 2 firewall.

But I usually get the message.
climateprediction.net - 2005-03-17 20:01:58 - Scheduler RPC to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded

It takes about 700 hours for my computers two finish one application each so I don't like delays.

---

This indicates a server problem. It should solve itself as the server load decreases, or whatever they do to fix it. It shouldn't be delaying your computations in any way, unless you are getting this when trying to download a new model.

This is quite an old problem and has been fixed in BOINC on client side, it now ignores everything outside oft the XML start/end tags. You must still be running 4.13 I guess.


From the checkin notes :

David 15 Oct 2004
- SCHEDULER_REPLY::parse(): tolerate extra lines
at start of scheduler reply

> This is quite an old problem and has been fixed in BOINC on client side, it
> now ignores everything outside oft the XML start/end tags. You must still be
> running 4.13 I guess.
>
>
> From the checkin notes :
>
> David 15 Oct 2004
> - SCHEDULER_REPLY::parse(): tolerate extra lines
> at start of scheduler reply

Yes, but that only made it into the development release of BOINC. It's <i>not</i> included in BOINC 4.19 but is in 4.25 (you'll still get the parse error but the scheduler request will succeed).

---

"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer

&gt; &gt; From the checkin notes :
&gt; &gt;
&gt; &gt; David 15 Oct 2004
&gt; &gt; - SCHEDULER_REPLY::parse(): tolerate extra lines
&gt; &gt; at start of scheduler reply
&gt;
&gt; Yes, but that only made it into the development release of BOINC. It's
&gt; <i>not</i> included in BOINC 4.19 but is in 4.25 (you'll still get the parse
&gt; error but the scheduler request will succeed).


The fix in parse() should not give any error/warnings anymore, if there's something outside of the scheduler_reply block. It starts the read loop with skipping all lines until it finds the scheduler_reply tag. It stops reading and returns when it finds the scheduler_reply end, ignoring everything behind.

I haven't seen a warning/error anymore for a long time so I assumed it was active in 4.19 already.

---

&gt; The fix in parse() should not give any error/warnings anymore, if there's
&gt; something outside of the scheduler_reply block. It starts the read loop with
&gt; skipping all lines until it finds the scheduler_reply tag. It stops reading
&gt; and returns when it finds the scheduler_reply end, ignoring everything
&gt; behind.

I stand corrected Ananas ;)

In my defence I was quoting from David Anderson's response when I raised this on the BOINC development mailing list the other day, but having looked at the BOINC code I concur that 4.25 won't generate the error :)

---

"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer

I was _very_ sure about the behaviour of this specific part of the code as I know it quite well ;-)

Greetings from Goettingen :-)

Volker