They know about and are working on the issue. From what I read, the database query speed is much slower than it should be, and they're working to speed it up.
They know about and are working on the issue. From what I read, the database query speed is much slower than it should be, and they're working to speed it up.
Maybe they need to know about Matt L's post over on SETI on an issue they found, and now are fixing/fixed for database slowness?
Anyway.. some good news. Bob may have gotten to the core of our database query problems, which is just in time as it seems other BOINC projects may be experiencing similar behavior as their databases swell. Basically it may just be as simple as adjusting database variables so specific queries won't keep whole tables in memory just in case updates are made during potentially long lookups. These queries don't require super-accurate information (like the result/workunit counts on the server status page), so if they are off by one or two out of a half million, so be it.
What I saw was Bruce Allen asking Dr. A about query speed issues, and getting a response which might have been helpful. There was code looking stuff there which I didn't understand.
The purpose of my post was to inform users that "yes, they know there's an issue". We/I haven't seen a post from them even acknowledging they know about it.
They know about and are
)
They know about and are working on the issue. From what I read, the database query speed is much slower than it should be, and they're working to speed it up.
RE: They know about and are
)
Maybe they need to know about Matt L's post over on SETI on an issue they found, and now are fixing/fixed for database slowness?
Anyway.. some good news. Bob may have gotten to the core of our database query problems, which is just in time as it seems other BOINC projects may be experiencing similar behavior as their databases swell. Basically it may just be as simple as adjusting database variables so specific queries won't keep whole tables in memory just in case updates are made during potentially long lookups. These queries don't require super-accurate information (like the result/workunit counts on the server status page), so if they are off by one or two out of a half million, so be it.
Wonder if this is the same issue?
What I saw was Bruce Allen
)
What I saw was Bruce Allen asking Dr. A about query speed issues, and getting a response which might have been helpful. There was code looking stuff there which I didn't understand.
The purpose of my post was to inform users that "yes, they know there's an issue". We/I haven't seen a post from them even acknowledging they know about it.
There is a very informal
)
There is a very informal thread from Gary Roberts from 12th Feb. about the problems.
Kind regards
Martin