We think we have found a simple fix for the database problems. The database has grown to 45 GB in size and has gotten too large for the physical memory of the machine that hosts it. However it turns out that due to some mistakes made in the project operations during the past weeks, about 80% of the work in the database is already completed. So we are running a db_purge task tonight that should remove this already-completed work from the database, leaving only the work-in-progress still in place in the database. If this is successful, then Einstein@Home should be back up and operating normally in about another 24 hours. Thank you for your patience!
45GB is overdoing it a LITTLE bit I'd say.
You're not thinking big enough. You can buy cots servers with as much as 256GB of ram. Properly configured the it'll run >$50k for hardware, and potentially several times that for software.
If you go beyond simple rackmount servers, IBM sells mainframes with upto 1.5TB of ram. For some reason IBM's not making it easy to see how much one of them costs. :D
One other item needs to be considered when the servers are put back on line, is the fact that some tasks that would have been reported in time are now past their deadline.
So I hope the Validators and any other process that re-issues because of missed deadlines are delayed, for at least 24 hours.
We think we have found a simple fix for the database problems. The database has grown to 45 GB in size and has gotten too large for the physical memory of the machine that hosts it. However it turns out that due to some mistakes made in the project operations during the past weeks, about 80% of the work in the database is already completed. So we are running a db_purge task tonight that should remove this already-completed work from the database, leaving only the work-in-progress still in place in the database. If this is successful, then Einstein@Home should be back up and operating normally in about another 24 hours. Thank you for your patience!
45GB is overdoing it a LITTLE bit I'd say.
You're not thinking big enough. You can buy cots servers with as much as 256GB of ram. Properly configured the it'll run >$50k for hardware, and potentially several times that for software.
If you go beyond simple rackmount servers, IBM sells mainframes with upto 1.5TB of ram. For some reason IBM's not making it easy to see how much one of them costs. :D
One other item needs to be considered when the servers are put back on line, is the fact that some tasks that would have been reported in time are now past their deadline.
So I hope the Validators and any other process that re-issues because of missed deadlines are delayed, for at least 24 hours.
If a unit gets re-issued because it wasn't returned before the deadline, and the original person returns it before the 'new' person, the original person gets the credit, as long as it is a valid crunch. I set my pc's to no new work, partly so I wouldn't be involved in all the day to day worries right now, but also to avoid this very problem. I do not want to get your past deadline unit, you return it and then I don't get credit for crunching it! After the Project is back up and running fine, I will come back. There are plenty of other projects out there that have short units, and need our help too.
You're forgetting the environment that BOINC is targeted to run in: science projects on a shoe-string budget.
And although Mainframes etc. don't need to have RAM > DB size because they generally run "proper" databases such as DB2 or Oracle, the "proper" databases cost $$$ more than MySQL.
But I'm pretty sure a small-ish IBM i5 could replace a large-RAM x86 server well enough and cheaply enough (you get DB2 for free with an i5).
You're forgetting the environment that BOINC is targeted to run in: science projects on a shoe-string budget.
And although Mainframes etc. don't need to have RAM > DB size because they generally run "proper" databases such as DB2 or Oracle, the "proper" databases cost $$$ more than MySQL.
But I'm pretty sure a small-ish IBM i5 could replace a large-RAM x86 server well enough and cheaply enough (you get DB2 for free with an i5).
Well, I guess BOINC server side code only works with MySQL (or does it have adaptors for other DBs???), but hey, as MySQL was bought by Sun and Sun was now bought by Oracle ... who knows where MySQL will go in the end :-)
You're forgetting the environment that BOINC is targeted to run in: science projects on a shoe-string budget.
And although Mainframes etc. don't need to have RAM > DB size because they generally run "proper" databases such as DB2 or Oracle, the "proper" databases cost $$$ more than MySQL.
But I'm pretty sure a small-ish IBM i5 could replace a large-RAM x86 server well enough and cheaply enough (you get DB2 for free with an i5).
Well, I guess BOINC server side code only works with MySQL (or does it have adaptors for other DBs???), but hey, as MySQL was bought by Sun and Sun was now bought by Oracle ... who knows where MySQL will go in the end :-)
CU
Bikeman
Some of the databases on Seti are Informix Matt Lebofsky post, so guess any SQL DB could be used.
RE: From the Home
)
You're not thinking big enough. You can buy cots servers with as much as 256GB of ram. Properly configured the it'll run >$50k for hardware, and potentially several times that for software.
http://configure.us.dell.com/dellstore/config.aspx?oc=becwwk1&c=us&l=en&s=bsd&cs=04&kc=category~rack_optimized
If you go beyond simple
)
If you go beyond simple rackmount servers, IBM sells mainframes with upto 1.5TB of ram. For some reason IBM's not making it easy to see how much one of them costs. :D
http://www-03.ibm.com/systems/z/hardware/z10ec/specifications.html
One other item needs to be
)
One other item needs to be considered when the servers are put back on line, is the fact that some tasks that would have been reported in time are now past their deadline.
So I hope the Validators and any other process that re-issues because of missed deadlines are delayed, for at least 24 hours.
RE: RE: From the Home
)
You're forgetting the environment that BOINC is targeted to run in: science projects on a shoe-string budget.
Seti Classic Final Total: 11446 WU.
RE: If you go beyond simple
)
I found 16 gigabytes of Server ram for $1000.00!
RE: One other item needs to
)
If a unit gets re-issued because it wasn't returned before the deadline, and the original person returns it before the 'new' person, the original person gets the credit, as long as it is a valid crunch. I set my pc's to no new work, partly so I wouldn't be involved in all the day to day worries right now, but also to avoid this very problem. I do not want to get your past deadline unit, you return it and then I don't get credit for crunching it! After the Project is back up and running fine, I will come back. There are plenty of other projects out there that have short units, and need our help too.
RE: You're forgetting the
)
And although Mainframes etc. don't need to have RAM > DB size because they generally run "proper" databases such as DB2 or Oracle, the "proper" databases cost $$$ more than MySQL.
But I'm pretty sure a small-ish IBM i5 could replace a large-RAM x86 server well enough and cheaply enough (you get DB2 for free with an i5).
RE: RE: You're
)
Well, I guess BOINC server side code only works with MySQL (or does it have adaptors for other DBs???), but hey, as MySQL was bought by Sun and Sun was now bought by Oracle ... who knows where MySQL will go in the end :-)
CU
Bikeman
RE: RE: RE: You're
)
Some of the databases on Seti are Informix Matt Lebofsky post, so guess any SQL DB could be used.
RE: ...I do not want to get
)
You most probably would get credit too!
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
