> On Aug 5, 2025, at 11:54 AM, Daniel Schwartz <d...@danielgschwartz.com> wrote:
>
> Hello again Rob,
>
> I just read that article you referenced by Cockroach Labs. I found the
> statement:
>
> "If we make the pool too small (i.e. choose too few connections), we’ll
> introduce latency, as operations have to wait for an available connection to
> open up before they can execute."
>
> This implies that if the number of concurrent accesses exceeds the number of
> connection objects in the pool, then the system simply waits until some of
> the currently active connections terminate and the connection objects become
> available.
>
> This is not what was happening with Glassfish. When the number of
> simultaneous requests exceeds the pool size, the system outputs an error
> message and crashes. It doesn't just wait until some active connections
> terminate so that the connection objects become available. It seems that
> different pooling systems have different policies.
>
Bottom posting is generally preferred. Crop to appropriate portion of original.
Are web crawlers making database requests?
Do you know which db calls are actually being made? And when?
Is the switch from 32 to 1000 the only change you made? I don’t think I am the
only one here who thinks that is a red herring.
You may need to give more information: db server, connection configuration,
servlet code, logs (app, middle tier, dbserver).
I suspect you may have been compromised. Is there anyone at your sight who can
comment on this?
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org
