Last night our sysman has installed on all our machines a new
sendmail.cf built from a newer Suse distribution. This cf contains
uncommented the line

O HostStatusDirectory=.hoststat

which in previous versions was commented out.

As a result my tripwire found new files in /var/spool/mqueue/.hoststat.
I discovered that it contained also older files (from about 2 years
ago when the system was originally installed on the machine ... the
system sendmail.cf was then immediately overwritten by our old
institute-wide sendmail.cf with HostStatusDirectory commented out, and
writing to the persistent database was suppressed.

I have read the man page and some other doc at sendmail.org about
hoststat and purgestat. Now my question are :

- do we gain something keeping the persistent database active ?
(So far we lived without it)

- will expired entries be purged automatically, or do we have to
run purgestat periodically (e.g. in a crontab ?)


--
----------------------------------------------------------------------
nospam@mi.iasf.cnr.it is a newsreading account used by more persons to
avoid unwanted spam. Any mail returning to this address will be rejected.
Users can disclose their e-mail address in the article if they wish so.

In news:Pine.LNX.4.61.0611090943490.23270@cbfrvqba.yn zoengr.vans.vg,
LC's NoSpam Newsreading account <nospam@mi.iasf.cnr.it> wrote:

> Last night our sysman has installed on all our machines a new
> sendmail.cf built from a newer Suse distribution. This cf contains
> uncommented the line
>
> O HostStatusDirectory=.hoststat
>
> which in previous versions was commented out.
....
> Now my question are :
>
> - do we gain something keeping the persistent database active ?

What did your "sysman" answer when you asked him?

In article <Pine.LNX.4.61.0611090943490.23270@cbfrvqba.ynzoen gr.vans.vg>
LC's NoSpam Newsreading account <nospam@mi.iasf.cnr.it> writes:
>
>I have read the man page and some other doc at sendmail.org about
>hoststat and purgestat. Now my question are :
>
>- do we gain something keeping the persistent database active ?
> (So far we lived without it)

Well, if everybody gained it would be used by default, wouldn't it?:-)
Whether you gain, only you can answer... If you've read the docs, you
presumably understand the benefit, i.e. queue runs can skip hosts that
have been found to be down by earlier queue runs, and thus complete
quicker and use less resources (plus you get the possibility to use the
SingleThreadDelivery option). Drawbacks include:

- Finding out that a host that has been down is up again may be delayed
by upto the Timeout.hoststatus setting (default 30 minutes) - and a
single connection failure will get a host "blacklisted".

- Checking the files increases the amount of I/O needed - on a busy
server the default location in the queue directory is probably the
worst possible choice.

So, if you often have a lot of queue runners sitting around trying to
connect to hosts that are down, delaying the processing of queued mail
that actually *can* be delivered, it may be a win - but there are other
ways to deal with that (e.g. queue groups, FallbackMXhost).

>- will expired entries be purged automatically, or do we have to
> run purgestat periodically (e.g. in a crontab ?)

AFAIK they will sit around forever unless manually purged - which
doesn't necessarily mean that you *have* to purge them...

--Per Hedeland
per@hedeland.org