Superzaffa’s Blog

Recently i had some troubles with maxdb. After a server restart the database would’nt open anymore. The error was a generic maxdb -24988 sql error

Looking in more deep in knldiag and knldiag.log i found

2011-06-15 11:37:15     0x134C     19615 DEVIO    Attaching devspace ‘D:\sapdb\DEV\sapdata\DISKD0001′
2011-06-15 11:37:15     0×1108 ERR 18177 DEVIO    No devspace on compressed file ‘D:\sapdb\DEV\sapdata\DISKD0001′
2011-06-15 11:37:15     0x134C ERR 53013 I/O      no devspaces on compressed file
2011-06-15 11:37:15     0x134C ERR 53017 I/O      DATADEV DEVNO: 1
2011-06-15 11:37:15     0x134C ERR 53016 I/O      D:\sapdb\DEV\sapdata\DISKD0001
2011-06-15 11:37:15     0x134C     19616 DEVIO    Detaching devspace ‘C:\sapdb\DEV\dbsys\SYS’
2011-06-15 11:37:15      0xCCC     19614 DBSTATE  I/O thread for ‘C:\sapdb\DEV\dbsys\SYS’ stopped
2011-06-15 11:37:15     0×1108     19618 DEVIO    Single I/O detach, ‘C:\sapdb\DEV\dbsys\SYS’, UKT:3
2011-06-15 11:37:15     0x134C     19651 CONNECT  Connection released, T5

And

2011-06-15 11:37:15     0×1108 ERR 18177 DEVIO    No devspace on compressed file ‘D:\sapdb\DEV\sapdata\DISKD0001′
2011-06-15 11:37:15     0x134C ERR 53013 I/O      no devspaces on compressed file
2011-06-15 11:37:15     0x134C ERR 53017 I/O      DATADEV DEVNO: 1
2011-06-15 11:37:15     0x134C ERR 53016 I/O      D:\sapdb\DEV\sapdata\DISKD0001

Poking around in forums (http://forums.sdn.sap.com/thread.jspa?threadID=590296) i found out that maybe my problems was related to datafiles compression. And that sounded strange to me because no one activated Filesystem compression on the volume of datafiles.

In fact the datafiles were compressed:

Then a colleague suggested the guilty: Disk Cleanup Utility. Disk . This nasty app should compress only unused files… i don’t know why MaxDb files falled in.

The faster way to uncompress such files was to move them back and forth to another server.

We experienced a funny problem with the interaction between Adobe Interactive Forms and amavis/clamav mail filter.

We have a SAP server sending out some Interactive PDF via mail but those mails get stuck in mail filter, we found in clamav logs:

PUA.Script.PDF.EmbeddedJS

It appears that if you activate PUA extensions on clamav, such filters can cause some false positive, including mail with Interactive PDF in attachment (the embedded JavaScript is the problem).

The fastest solution we found to this problem was to disable PUA extension for Script, adding this line:

ExcludePUA Script

to clamd.conf.

For reference:

http://www.clamav.net/lang/en/support/faq/pua/

Maybe a better idea would be to create a correct signature for the false positive and insert it into a  local.ign2 file for whitelist this kind of attachments.

Ref: http://www.clamav.net/doc/latest/signatures.pdf

An example of permission set up on a nexenta share (TEST):

/usr/sun/bin/chmod -R  A+group:2147483653:rwxpdDaARWcCos:fd—–:allow TEST
/usr/sun/bin/chmod -R  A+group:2147483651:r-x—a-R-c–s:fd—–:allow TEST
/usr/sun/bin/chmod -R  A+user:2147483648:rwxp–aARWcCos:fdi—-:allow TEST
/usr/sun/bin/chmod -R  A+group:2147483648:rwxpdDaARWcCos:fd—–:allow TEST
/usr/sun/bin/chmod -R  A+group:2147483652:r-x—a-R-c–s:fd—–:allow TEST
/usr/sun/bin/chmod -R  A+group:2147483650:rwxp–aARWcCos:fd—–:allow TEST

To view such permissions:

/usr/sun/bin/ls -V

Obviously lauch such commands from expert mode console.

In these days i’m playing a lot with nexenta tools.

Hi have a working server builded on:

• HP DL 180 G6 20Gb ram
• 2×80 Gb sata disk for OS (mirror pool)
• 6x750Gb sata disk  in al zpool (2 groups,3 disks raidz1)

all those 8 disk are on the server backplane. Then we added an LSI HBA and we attached a SUN J4400 JBOD filled with 24 1Tb disks (mirror pool configuration for maximum performances)

All went well until today: a disk form syspool went offline. Luckily i had some spare disks (same type), and i tryed to put it back online. I tought it was an easy operation as i experienced on other zpools where i changed failed disks.

But this time, maybe because this is the syspool, the operation was quite different.

First: no operation from web console. It seems that it is impossibile to manage syspool luns from web interface.

Operations

• insert disk
• format
• fdisk -> delete all partitions, create only one 100% Solaris2
• back to format menu
• partition
• build a “slice” partition like the other disk in the mirror
• write the label
• quit partition and format
• zpool replace -f syspool c0t1d0s0 c0t1d0s0  (it will replace faulty c0t1d0s0 with good one)
• wait until resilver finish
• cd /boot/grub
• installgrub stage1 stage2 c0t1d0s0

Et voilà (kudos to http://dd-b.net/ddbcms/2009/01/opensolaris-zfs-root-pool-mirroring/)

It seems that the new security policy of Adobe Flash Player brings some troubles to BO reporting

The problem lies in crossdomain scripting. You need to explicitly accept all connection from everywhere (or set up something more accurate). By default Adobe policy is very restrictive and blocks any interaction.

For the crossdomain.xml generation, take a look at SAP note

1201370 – Error #2170 when previewing or exporting to Business Objects Enterprise an Xcelsius dashboard