For those of you that still are running CentOS 4... I have one system that is still going... there is a problem with the newest release of Dovecot under mbox. Certain spam is causing this error when users try to log on.
file lib.c: line 37 (nearest_power): assertion failed: (num <= ((size_t)1 << (BITS_IN_SIZE_T-1)))
Rolling back to a previous release fixes these issues. I'm not bothering to file a bug with Redhat as the EOL is rapidly approaching and I just about have my one system's users moved to a new server.
I have not as of yet seen this problem on CentOS 5 mbox systems, but I don't have many users on those systems either as I'm 'slowly' migrating all to CentOS 6 Maildir systems.
On Sun, Oct 23, 2011 at 10:19 AM, John Hinton webmaster@ew3d.com wrote:
For those of you that still are running CentOS 4... I have one system that is still going... there is a problem with the newest release of Dovecot under mbox. Certain spam is causing this error when users try to log on.
file lib.c: line 37 (nearest_power): assertion failed: (num <= ((size_t)1 << (BITS_IN_SIZE_T-1)))
Rolling back to a previous release fixes these issues. I'm not bothering to file a bug with Redhat as the EOL is rapidly approaching and I just about have my one system's users moved to a new server.
I have not as of yet seen this problem on CentOS 5 mbox systems, but I don't have many users on those systems either as I'm 'slowly' migrating all to CentOS 6 Maildir systems.
I got bit by this bug as well. I rolled back to dovecot-0.9.11-9 for now until I find time to upgrade to CentOS 5 or 6.
on 11/1/2011 10:30 AM Grant McChesney spake the following:
On Sun, Oct 23, 2011 at 10:19 AM, John Hintonwebmaster@ew3d.com wrote:
For those of you that still are running CentOS 4... I have one system that is still going... there is a problem with the newest release of Dovecot under mbox. Certain spam is causing this error when users try to log on.
file lib.c: line 37 (nearest_power): assertion failed: (num<= ((size_t)1<< (BITS_IN_SIZE_T-1)))
Rolling back to a previous release fixes these issues. I'm not bothering to file a bug with Redhat as the EOL is rapidly approaching and I just about have my one system's users moved to a new server.
I have not as of yet seen this problem on CentOS 5 mbox systems, but I don't have many users on those systems either as I'm 'slowly' migrating all to CentOS 6 Maildir systems.
I got bit by this bug as well. I rolled back to dovecot-0.9.11-9 for now until I find time to upgrade to CentOS 5 or 6.
DO yourself a favor and use a dovecot from a third party repo... the 0.9 series is YEARS old.
On 11/1/2011 3:53 PM, Scott Silva wrote:
on 11/1/2011 10:30 AM Grant McChesney spake the following:
On Sun, Oct 23, 2011 at 10:19 AM, John Hintonwebmaster@ew3d.com wrote:
For those of you that still are running CentOS 4... I have one system that is still going... there is a problem with the newest release of Dovecot under mbox. Certain spam is causing this error when users try to log on.
file lib.c: line 37 (nearest_power): assertion failed: (num<= ((size_t)1<< (BITS_IN_SIZE_T-1)))
Rolling back to a previous release fixes these issues. I'm not bothering to file a bug with Redhat as the EOL is rapidly approaching and I just about have my one system's users moved to a new server.
I have not as of yet seen this problem on CentOS 5 mbox systems, but I don't have many users on those systems either as I'm 'slowly' migrating all to CentOS 6 Maildir systems.
I got bit by this bug as well. I rolled back to dovecot-0.9.11-9 for now until I find time to upgrade to CentOS 5 or 6.
DO yourself a favor and use a dovecot from a third party repo... the 0.9 series is YEARS old.
We've dealt with if for nearly 7 years now and only have a few months to go. The problems have been few. I posted this to help those make it through to February. I delayed moving folks from the v4 systems waiting for the v6 systems so as to gain a couple or few more years before EoL for them and for other reasons that v5 does not properly address. Clients don't like to be moved around. In a perfect world, email client programs would not have problems with these moves... but we don't live in a perfect world. Those problems irritate the clients and increase our tech support by multiple times. Heading off into a repo 1.x upgrade at this point is rather silly IMO.
-
Grant McChesney -
John Hinton -
Scott Silva