Bug ID 756094: DNS express in restart loop, 'Error writing scratch database' in ltm log

Last Modified: May 29, 2024

Affected Product(s):
BIG-IP LTM(all modules)

Known Affected Versions:
14.1.0, 12.1.4

Fixed In:
15.0.0, 14.1.0.2, 13.1.1.5, 12.1.4.1

Opened: Jan 20, 2019

Severity: 2-Critical

Symptoms

DNS express (zxfrd) daemon gets stuck in a restart loop with the messages: -- In /var/log/ltm: Error writing scratch database (no error information available), serving database is unchanged. zxfrd will exit and restart. -- The system posts the following message on the command line every few seconds: emerg logger: Re-starting zxfrd

Impact

Zone updates from the DNS master servers are not processed.

Conditions

An update to an SOA record (and only an SOA) is received through a incremental zone transfer update (IXFR).

Workaround

As a partial workaround, the DNS express cache files can be removed, forcing zxfrd to pull the entire zone using an AXFR request. To do so, use the following commands, in sequence: bigstart stop zxfrd rm /shared/zxfrd/* bigstart start zxfrd Note: DNS express will not be able to service DNS responses until the zone transfers have completed. For this reason, this procedure should be carried out on the standby device, if possible.

Fix Information

The system now properly handles IXFRs that contain only starting and ending SOA RRs, and no other RRs.

Behavior Change

Guides & references

K10134038: F5 Bug Tracker Filter Names and Tips