- To: email@example.com
- Subject: SMTP mailjam!?
- From: firstname.lastname@example.org (Lee Ziegenhals)
- Date: Wed, 22 Jan 92 9:05:12 CST
>!) Several local users of gerards distribution .exe (1.8b) have
> had problems with severe "terminal mailjam".
>I relay some of this feed to local amprnet users.. when i
>returned from a brief trip to the great white north i had
>56 queued messages that caused gri-NOS to lockup the first
>time the SMTP timer kicked. I finally had to nuke the
>entire queue to get the pipes unclogged!
I thought I would add a bit here, as I am one of the "local users"
We have a node here that is a 12MHz 286 with a 30 MB hard drive
and a "not so fast" disk controller. When the SMTP queue starts
getting large, perhaps more than 30 or 40 messages, things start
to get very slow. The symptom is that, whenever the mqueue directory
is accessed, NOS spends an enourmous amount of time accessing the
disk. During this time, the disk light is flickering madly, the
head is bouncing all over the place, and the response to commands
is extremely slow. Depending on the size of the queue, this will
go on for up to 3 or 4 minutes.
This happens whenever the smtp timer expires, or even when a message
is successfully delivered and is deleted from the queue. It seems
to occur primarily when most of the mail is queued to a single
system, though I have not tested that theory.
A disk cache helps, but does not solve the problem. The disk could
probably stand to be de-fragmented, though.
I thought this was an artifact of my hardware until Jack mentioned
the same problem.