Looking for some help as we've been investigating this for a while
with no resolution.

This has been occurring in a somewhat random pattern on a couple of
our customer sites. When it happens it is always at about the same
time of day, but it doesn't happen every day. Sometimes weeks go by
between occurrences, then it will happen a few days in a row.
Characteristics are as follows:

- Windows performance log shows active requests suddenly rise to above
the normal numbers (i.e. from 2 or 3 to 13) while unscheduled requests
remain 0 for about 10 minutes, then unscheduled requests spike and
stay high (i.e. 22) until the server is rebooted.

- At the time of the initial rise in active requests, the database can
no longer be connected to. ODBC message is "Found server but a
communication error occurred"

- The database service can't be stopped and dbmlsrv8.exe can't be
terminated

- The lockups always occur while daily dbbackup of one of the
databases is running from a backup server. (four databases are being
served and backed up daily. Three of these are 5GB to 8GB in size;
backup normally takes 30 - 60 minutes each)

- Nothing unusual noted in -z output. Now running -zr all, but only
one occurrence captured and nothing unusual noted so far.

- dbmlsrv8 not using unusual amounts of CPU or RAM, nether is any
other process in task manager

- Lockups happen during a time of low user activity

Some system characteristics:
- Servers are Windows 2003
- Backup server is on the same LAN as the server
- Servers are Mobilink remotes with dbmlsync running every 5 minutes
- There are services running that use the databases regularly
throughout the day

Any ideas or suggestions would be most welcome.

Thanks,
Bob Leviton

Don't you mean 8.0.3.5267?

The following is my UNRELIABLE memory talking: Historically there have
been some nasty bugs involving backups and checkpoints and the like,
with hangup being the common symptom. These issues came and went, over
the years, ending with later EBFs of 9.0.2. One way to trigger one of
these bugs was to have some other connection issue a CHECKPOINT while
a backup is running (and suppressing checkpoints).

....again, all that is from an old guy's memory.

And 8.0.3.5267 is very old. There are later EBFs of V8 (I am running
8.0.3.5379 for legacy testing) but V8 is no long supported... that
means there will be no more fixes.

So, try a later EBF, and remove explicit CHECKPOINT commands from your
applications if they exist, but even better, move forward to V11.

Breck


On Thu, 29 Oct 2009 11:57:19 -0700 (PDT), Bob Leviton
<leviton.r (AT) emssatcom (DOT) com> wrote:

--
Breck Carter http://sqlanywhere.blogspot.com/

RisingRoad SQL Anywhere and MobiLink Professional Services
breck.carter (AT) risingroad (DOT) com

On Oct 30, 8:35*am, "Breck Carter [TeamSybase]"
<NOSPAM__breck.car... (AT) gmail (DOT) com> wrote:
Breck,

Sorry, yes, it's 8.0.3.5267. I suspected something like you describe.
In the most recent instance I saw an automatic checkpoint issued for
one of the databases (not the one being backed up) just after the
start of the backup and just before the lockup.

Thanks,
Bob