Mailing List ecs-isp@2rosenthals.com Archived Message #779 previous message next message back to list

From: "Paul Smedley" <ecs-isp@2rosenthals.com> Full Headers
Undecoded message
Subject: Re: [eCS-ISP] Bind 9.11.37 issue - ticket #784 - update
Date: Wed, 3 Jul 2024 19:17:03 +0930
To: eCS ISP Mailing List <ecs-isp@2rosenthals.com>

Hi folks,

On 3/7/24 08:09, Steven Levine wrote:
In <list-10080011@2rosenthals.com>, on 06/28/24
    at 10:16 AM, "Massimo S." <ecs-isp@2rosenthals.com> said:

Hi Massimo,

i add, this build start at 92-100% cpu and it's completely unresponsive
can't be killed and nslookup don't work
it don't answer to any request, neither in the first seconds after the
start
Too bad.  I suspect the timeout call that Paul implemented is not quite
correct.  My thinking was that a 10 second timeout would be a good trade
off between responsiveness and the ability to kill named.

I'd have to see a copy of Paul's patch to have a useful comments on what
Paul implmented.

I chose 5 seconds...

#elif defined(USE_SELECT)
            LOCK(&manager->lock);
            memmove(manager->read_fds_copy, manager->read_fds,
                manager->fd_bufsize);
            memmove(manager->write_fds_copy, manager->write_fds,
                manager->fd_bufsize);
            maxfd = manager->maxfd + 1;
            UNLOCK(&manager->lock);

#ifndef __OS2__
            cc = select(maxfd, manager->read_fds_copy,
                    manager->write_fds_copy, NULL, NULL);
#else
            struct timeval  tv;
            /* Wait up to five seconds. */

            tv.tv_sec = 5;
            tv.tv_usec = 0;
            cc = select(maxfd, manager->read_fds_copy,
                    manager->write_fds_copy, NULL, &tv);
#endif /* __OS2__ */
Subscribe: Feed, Digest, Index.
Unsubscribe
Mail to ListMaster