firebird-devel

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Dimitry S. <sd...@ib...>]

27.01.2011 11:54, Alex Peshkoff wrote:
> Certainly it's virtual space, but anyway - taking into an account that
> CPUs became a bit faster since IB6, it's reasonable to think about
> making timeout smaller.

   CPUs became faster, but Windows thread and memory management - don't.

-- 
   SY, SD.

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Adriano d. S. F. <adr...@gm...>]

On 27/01/2011 09:02, Dimitry Sibiryakov wrote:
> 27.01.2011 11:54, Alex Peshkoff wrote:
>> Certainly it's virtual space, but anyway - taking into an account that
>> CPUs became a bit faster since IB6, it's reasonable to think about
>> making timeout smaller.
>     CPUs became faster, but Windows thread and memory management - don't.
>
AFAIR, since Vista, Windows has thread pool API...

Anyway, I don't think we should use such non-standard API that requires 
very different implementation for others platforms.


Adriano

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Vlad K. <hv...@us...>]

> 27.01.2011 13:28, Dimitry Sibiryakov wrote:
> 
>> Exactly to make threads' creation/deletion seldom. Sleeping thread consumes no resources.
> 
> Doesn't 2MB of stack count, especially when you have hundreds of them?

    Dealing with 'out of memory' conditions - yes, counts. Especially for 32-bit process.

    Imagine application server wich creates own pool of connections at startup. It easy could 
force Firebird to create 100th of threads for a short period of time. Then app server start to serve 
users and some of them requires much memory. But we can't allocate it just because we have 
100th idle worker threads wasting 100th of MBs of virtual adress space. 

    Just dumb example of course :)

Regards,
Vlad

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Vlad K. <hv...@us...>]

> There is a 30MB Database, should I zip it and email it to you? We can also
> give you the test query and test program as well. 

    Database, queries, program - all what is necessary to reproduce the issue.
And don't mail it to me, better make it available for download somewhere.

Regards,
Vlad

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Yi Lu <yl...@ra...>]

I have sent to you a link. 
-- 
View this message in context: http://firebird.1100200.n4.nabble.com/32-bit-Firebird-attempting-to-create-large-amount-of-thread-on-64-bit-server-tp3236747p3242643.html
Sent from the firebird-devel mailing list archive at Nabble.com.

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Vlad K. <hv...@us...>]

> it would be interesting to know why there
> is a behavior change between FB 2.5 and FB 2.1. 

    Am i already asked for the test case ?

Regards,
Vlad

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Yi Lu <yl...@ra...>]

There is a 30MB Database, should I zip it and email it to you? We can also
give you the test query and test program as well. 
-- 
View this message in context: http://firebird.1100200.n4.nabble.com/32-bit-Firebird-attempting-to-create-large-amount-of-thread-on-64-bit-server-tp3236747p3242494.html
Sent from the firebird-devel mailing list archive at Nabble.com.

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Alex P. <pes...@ma...>]

 On 01/27/11 14:02, Vlad Khorsun wrote:
>> Certainly it's virtual space, but anyway - taking into an account that
>> CPUs became a bit faster since IB6, it's reasonable to think about
>> making timeout smaller.
>     I'd said timeout should be tuned on the fly dependent on real load.

If we will have an ability to ask subsystems 'free some memory if
possible', this will be great.

> Anyway i don't see this as "issue" which requires ugrent fix.
>

Yes, certainly.

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Yi Lu <yl...@ra...>]

This crash issue has prevented us from using FB 2.5 32-bit in our product on
high load 32-bit systems. Since FB 2.1 does not create so many threads in
identical circumstances, it is more robust for a high-load 32-bit operating
system. While this issue will become less of a problem as more people move
towards 64-bit operating systems, it would be interesting to know why there
is a behavior change between FB 2.5 and FB 2.1. 
-- 
View this message in context: http://firebird.1100200.n4.nabble.com/32-bit-Firebird-attempting-to-create-large-amount-of-thread-on-64-bit-server-tp3236747p3242421.html
Sent from the firebird-devel mailing list archive at Nabble.com.

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Dimitry S. <sd...@ib...>]

26.01.2011 16:32, Yi Lu wrote:
>
> 3 clients, each with 99 threads and each client thread sending a query to
> server.

   What protocol? XNET, Named Pipes, TCP?

-- 
   SY, SD.

Re: [Firebird-devel] 32-bit Firebird attempting to create large amount of thread on 64-bit server

[Yi Lu <yl...@ra...>]

It was connecting through TCP.
-- 
View this message in context: http://firebird.1100200.n4.nabble.com/32-bit-Firebird-attempting-to-create-large-amount-of-thread-on-64-bit-server-tp3236747p3238325.html
Sent from the firebird-devel mailing list archive at Nabble.com.