status of solaris-10 zimbra port

[serivo]

I'd like to post my impression from first days of zimbra 5.0.15 (32 bit) running on solaris-10.
First, it successfully upgraded zimbra-4.5.4 made by dijichi2.
Immediately after upgrade I did server name change, and it also worked.
Then, recently I run stress test on the system, by smtp-source, both to the port 25 and to the 7025 speaking LMTP, and have some 200 thousands messages, each 16k size, delivered to test mailbox. Imaptest (from dovecot sources) also run successfully, at least not less successfully as it runs on linux zimbra installations.

Problems I have seen:
1. zmmtaconfig was killing zimbra java applications and was trying to restart it, I did not yet understand why and I disabled this functionality by setting zmmtaconfig_enable_config_restarts=FALSE
2. Sys::Syslog perl module is broken in Solaris, see Bug ID: 6503341 Sys::Syslog is broken in Solaris 10, and as a workaround I first added `/usr/bin/logger -t "$ident\[$$\]" -p $facility.info "$$:$level: $msg"`; into zimbramon/lib/Zimbra/Mon/Logger.pm, but then installed recent Sys::Syslog module from CPAN.
3. I needed to install perl modules IO::Socket::INET6 and Socket6, for amavis, even if ipv6 is not enabled on the system.
4. Still have not get graphs working for mta, - I have only amavis being graphed, zmlogprocess finds 0 records for all but amavis.

--
Sergey.

[serivo]

On one fresh installation of zimbra 5.0.15 on Solaris I was struggling for a long time with "Could not reserve enough space for object heap".
Finally I've found
"As a starting point for a server based on a single JVM, consider setting the maximum heap size to 1/4 the total physical memory on the server and setting the minimum to 1/2 of the maximum heap. Sun recommends that ms be set to somewhere between 1/10 and 1/4 of the mx setting. They do not recommend setting ms and mx to be the same. Bigger is not always better for heap size. In general increasing the size of the Java heap improves throughput to the point where the heap no longer resides in physical memory. Once the heap begins swapping to disk, Java performance drastically suffers. Therefore, the mx heap setting should be set small enough to contain the heap within physical memory. "
The patch to resolve this was:
---
diff --git a/FRANKLIN/ZimbraServer/src/bin/zmmailboxdctl b/FRANKLIN/ZimbraServer/src/bin/zmmailboxdctl
index 3dad803..798aab7 100755
--- a/FRANKLIN/ZimbraServer/src/bin/zmmailboxdctl
+++ b/FRANKLIN/ZimbraServer/src/bin/zmmailboxdctl
@@ -35,6 +35,7 @@ zmassert -d ${mailboxd_directory}
#
memkb=$(zmsysmemkb)
jm=`$EXPR $memkb '*' ${mailboxd_java_heap_memory_percent} / 100 / 1024`
+jmm=`$EXPR $jm / 2`

#
# Expand war files.
@@ -101,7 +102,7 @@ case "$1" in
fi
echo -n "Starting mailboxd..."
/opt/csw/bin/sudo ${zimbra_home}/libexec/zmmailboxdmgr start \
- -Xms${jm}m -Xmx${jm}m ${mailboxd_java_options} < /dev/null > /dev/null 2>&1
+ -Xms${jmm}m -Xmx${jm}m ${mailboxd_java_options} < /dev/null > /dev/null 2>&1
status=$?
if [ $status = 0 ]; then
echo "done."
---