Any idea when the export of Seti credit totals to the likes of BoincStats is going to kick in? I'm over 20k adrift (even between the two figures for Seti on my seti account page) and that was before I left for work.

Rob.

---

They published something last night, but that does not mean that it is up to date stats.

http://setiathome.berkeley.edu/stats/

---

They published something last night, but that does not mean that it is up to date stats.

http://setiathome.berkeley.edu/stats/

Did they? My values show zero for yesterday and have not moved yet today (and that's on my SETI Boing account page).

No panic but looking at BoincStats filtered for SETI everyone is showing zero, so for once it's not just me.

Rob.

---

I have a similar thread going in "Questions and Answers : Web site"

The problem is not just for you, it's for everybody, teams, users.. all. (our stats does not include the hosts, but I suspect that one has the same issues..)

I think the cause is in the MySql problems Berkeley is having.. It looks like the export script is doing it's work fine, but is exporting from a "frozen table/database"..

---

The SETI@Home Gauntlet 2012 april 16 - 30| info / chat | STATS

IIRC, the stats export is run from the Replica DB. Maybe it hasn't been updated due to all the DB issues.

I've got almost 40K "missing". ;)

---

Calm Chaos Forum...Join Calm Chaos Now

IIRC, the stats export is run from the Replica DB. Maybe it hasn't been updated due to all the DB issues.

I've got almost 40K "missing". ;)

Thats fine, I'm sure it will catch up eventualy. It's not a problem, I'm just curious as to what is going on. Nice to have an up-front answer.

Thanks.

Rob.

---

Another day with zero credits exported. Hey-Ho!

---

Another day with zero credits exported. Hey-Ho!

Stats are exported from the replica database to limit loading, and from all other reports they had a very rocky weekend.

The stats are exported once per day (and picked up by the stat sites once per day). If that doesn't work tonight, I'm sure they'll be digging into it (unless they know the replica is bad and turned it off on purpose).

---

IIRC, the stats export is run from the Replica DB. Maybe it hasn't been updated due to all the DB issues.

I've got almost 40K "missing". ;)

Thats fine, I'm sure it will catch up eventualy. It's not a problem, I'm just curious as to what is going on. Nice to have an up-front answer.

Thanks.

Rob.

See this post.

---

The new stats update is there, thanx for fixing this (also)!

---

The SETI@Home Gauntlet 2012 april 16 - 30| info / chat | STATS

Question,

Yesterday the data export was at 9:31:19 UTC, today the export was generated/finished at 09:38:18 UTC.

Usually the exports were around midnight UTC. So the difference is rather big..

When I look at the stats-folder I see that the tables.xml has the date: 28-Apr-2009 02:38

It almost looks like the "export server" (does that one exist?) is running in the wrong time-zone or something?


PS: Yesterday after the data sync, the accumulated credits for most teams/users etc was more then one day, so the exported data is "in sync", just the moment seems te have gotten out of sync..

---

The SETI@Home Gauntlet 2012 april 16 - 30| info / chat | STATS

Question,

Yesterday the data export was at 9:31:19 UTC, today the export was generated/finished at 09:38:18 UTC.

Usually the exports were around midnight UTC. So the difference is rather big..

When I look at the stats-folder I see that the tables.xml has the date: 28-Apr-2009 02:38

It almost looks like the "export server" (does that one exist?) is running in the wrong time-zone or something?


PS: Yesterday after the data sync, the accumulated credits for most teams/users etc was more then one day, so the exported data is "in sync", just the moment seems te have gotten out of sync..

Berkeley is currently -7 UTC so that would be 02:38 PDT, the stats server does not have UTC on it.

---

OK.. but that changed then! I have (re)built a stats-system for my team. This went live at 28th of February. All that time the data export was created around or mostly a bit after midnight UTC. That seems to have changed to around 9:30 UTC since yesterday.

Not that I see this as a problem, I just thought it was odd, since there does not seem to be a reason in my view to change the export moment.

PS: I must confess that midnight UTC is preferable to me/our server since the data synchronisation takes place at night then (Dutch time = UTC+2 ;)

---

The SETI@Home Gauntlet 2012 april 16 - 30| info / chat | STATS

And got fixed again? ;) Last export: 22:16:03 28-04-2009 UTC (so the stats-sites won't show you had a great production day, because the interval was a bit more then half a day ;)

Some questions are floating around in my head..:

• Is this the right sub-forum to post these kind of questions/remarks?
  
• Is it appreciated that I mention/ask these kind of discontinuities?
  
• Can one think (realisticly ;) that this got fixed, but "The Berkeley People" simply are to bussy to mention it?
  

---

The SETI@Home Gauntlet 2012 april 16 - 30| info / chat | STATS

And got fixed again? ;) Last export: 22:16:03 28-04-2009 UTC (so the stats-sites won't show you had a great production day, because the interval was a bit more then half a day ;)

Some questions are floating around in my head..:

• Is this the right sub-forum to post these kind of questions/remarks?
  
• Is it appreciated that I mention/ask these kind of discontinuities?
  
• Can one think (realisticly ;) that this got fixed, but "The Berkeley People" simply are to bussy to mention it?
  

I think Matt actually did mention it.

---

OK, I'll keep mentioning oddities.. (it's nice to think one contributes.. and then I don't just mean my RAC ;)

The first one.. and this has been asked before.. the 'number-crunching' section is the most active.. but I have in the past posted in 'Questions and Answers : Web site' too.. what is a good spot?

---

The SETI@Home Gauntlet 2012 april 16 - 30| info / chat | STATS

OK, I'll keep mentioning oddities.. (it's nice to think one contributes.. and then I don't just mean my RAC ;)

The first one.. and this has been asked before.. the 'number-crunching' section is the most active.. but I have in the past posted in 'Questions and Answers : Web site' too.. what is a good spot?

Good question. I don't think the "Technical News" is the right place, although I think if you keep up with that section that most operational glitches get mentioned there sooner or later if there is anything worth saying.

We know from reading the technical news that the stats come from the backup (replica) database to ease loading on the primary.

We also know that when there is a problem in general, the stat updates are usually fixed last -- and we know that they often recopy the database during an outage instead of letting the normal "sync" process deal with it.

Q&A is manned with volunteers like us who may not know any more than we can observe. I don't think it's wrong to ask.

I've noted that questions asked here are often answered (more or less) by Matt in his (mostly-daily) technical news posts.

---

OK.. but that changed then! I have (re)built a stats-system for my team. This went live at 28th of February. All that time the data export was created around or mostly a bit after midnight UTC. That seems to have changed to around 9:30 UTC since yesterday.

Not that I see this as a problem, I just thought it was odd, since there does not seem to be a reason in my view to change the export moment.

PS: I must confess that midnight UTC is preferable to me/our server since the data synchronisation takes place at night then (Dutch time = UTC+2 ;)

Wearing my developer's hat for a moment, I'd probably code my app. so it checked for an update (file timestamp changed) every couple of hours or so, and if it downloaded an update, wait something like 23 hours before checking again.

Changing that from two hours to 12 hours after a few tries would probably be more server-friendly.

That way it'd tend to download near an update but not load the connection much.

---

Wearing my developer's hat for a moment, I'd probably code my app. so it checked for an update (file timestamp changed) every couple of hours or so, and if it downloaded an update, wait something like 23 hours before checking again.

Ahhh.. but for all those stats junkies out there an update is an update.. and they want it ;)

Changing that from two hours to 12 hours after a few tries would probably be more server-friendly.

That way it'd tend to download near an update but not load the connection much.

Ah.. the tables.xml is only a few hundred bytes ;)

And "all" those stats-sites do alleviate the stress on the Berkeley-site! I am certain that the "burden of exporting" is more then compensated by this..

---

The SETI@Home Gauntlet 2012 april 16 - 30| info / chat | STATS

Wearing my developer's hat for a moment, I'd probably code my app. so it checked for an update (file timestamp changed) every couple of hours or so, and if it downloaded an update, wait something like 23 hours before checking again.

Ahhh.. but for all those stats junkies out there an update is an update.. and they want it ;)

Changing that from two hours to 12 hours after a few tries would probably be more server-friendly.

That way it'd tend to download near an update but not load the connection much.

Ah.. the tables.xml is only a few hundred bytes ;)

And "all" those stats-sites do alleviate the stress on the Berkeley-site! I am certain that the "burden of exporting" is more then compensated by this..

My understanding is that the export files are created once per day, so the goal would be to download just as soon after they are created as possible.

... and I'd want to do so without having to monitor the servers for any schedule changes at Berkeley.

I understand that some of the XML files are quite large, but it's been a long time and I no longer remember where they can be downloaded.

---