[Date Prev][Date Next][Thread Prev][Thread Next][Interchange by date
][Interchange by thread
]
[ic] Verisign PFPRO Transactions Getting Dropped - SOLVED!
G'Day Ron Again,
Two other things you may want to be aware of Ron...
01) When we were having decussins with Signio/VeriSign they told us flat out
that the timeout should NEVER be set below 30... We originally ran it with 45
and it made NO difference.
02) You may want to also check your available RAM, swapping, and load factor.
All contribute to slow creation of tmp files.
This problem does go away by resolving IO, Swapping, & tmp file problems.
John Beima
Quoting Ron Phipps <rphipps@reliant-solutions.com>:
> > From: interchange-users-admin@interchange.redhat.com
> [mailto:interchange-
> > users-admin@interchange.redhat.com] On Behalf Of John Beima
> >
> > Boy am I tired of seeing this one. (-:
>
> You can't be that tired of it... it doesn't show up as often as "What is
> the admin password" :)
>
> >
> > This problem is cause because the IO on your box is to slow... It is
> the
> > same
> > thing with the PGP errors.
> >
> > Interchange uses temp files back and forth for pgp encrypting. PF Pro
> > seems to
> > use temp files as well... When your IO channel gets a little bogged
> down,
> > it
> > can't create the files fast enough for the programs, so it appears to
> > Interchange like the transaction or the encryption has failed.
>
> The new Signio.pm does use temp files. But the old one does not and we
> tested with that module as well and saw the same problem.
>
> > Simple solution, repace your IO controllers and hard drives with
> faster
> > PROPER
> > drives... Anyone that thinks ATA100 or ATA133 is anything near the
> speed
> > of SCSI
> > is dreaming... A standard ATA100 drive can deal with about 500 IOs per
> > second
> > and ATA RAID makes NO difference... A regual UltraWide SCSI can deal
> with
> > 3000,
> > SCSI RAID gets to even around 5000... Not to mention on the SCSI chain
> > EVERY
> > device can run at once, where as on the IDE chain only one can.
>
> Unfortunately the server in question is running U2W SCSI drives and it
> won't be as easy as throwing some new hardware at it.
>
> > You may also wish to check your CPU load factor, but most likely you
> wont
> > see it
> > here. You can try a "vmstat 5" command and watch your "bi" & "bo".
> This
> > will
> > give you a closer look at how back logged your box really is.
>
> Thanks for the debugging command... I'll look more into this and
> hopefully it will give us an insight into what is happening.
>
> > Simply put, if you have a busy site, you need a REAL server, not some
> > Pentium
> > III desktop with Unix on it. Tis the cost of doing business.
>
> Agreed, this server should be more then plenty for a single catalog that
> is not heavily hit.
>
> The mystery continues ;)
> -Ron
>
> _______________________________________________
> interchange-users mailing list
> interchange-users@interchange.redhat.com
> http://interchange.redhat.com/mailman/listinfo/interchange-users
>
John Beima
jbeima@palb.com, support@alocalagent.com, and support@alocalchurch.com
P.A.L.B. Systems - Phone: (780)451-1086 - Fax: (780)447-4760
11639-122 Street, Edmonton, Alberta, Canada, T5M 0B6
Affordable Web Pages - Phone: (888)932-9990 - Fax: (256)351-7297
2713B Spring Place SW, Decatur, Alabama, United States, 35603