Virgo Logbook

Computers Network (Name Server activity)

cortese - 13:27 Friday 25 September 2020 (49498)

Cm connection problem triggered by DNS reverse resolution unavailability

Today the DNS reverse resolution providing the DNS name of hosts, knowing their IP address, for the Cascina machines was broken because of a consistency problem of the zone files which affected all the DNS servers.

This problem has shown a, somewhat neglected, dependency of the Cm namserver operation on the DNS reverse resolution since no new cm connections are accepted if the Cm nameserver cannot obtain the DNS name of the client from its IP address.

The problem caused also the progressive unavailability of many rtpcs.

This kind of DNS problems often go unnoticed for a number of days due to the caching mechanisms of the clients. Therefore we will put in place a specific alarming to check the health of the Cascina reverse DNS zones.

Comments to this report:

masserot - 14:19 Friday 25 September 2020 (49500)

This was not only Cm connection problems. The rtpcs lost too theirs accesses through NFS to the Virgo's disks

Now everithing is back and all the loops(SBE, LC) are closed.

cortese - 16:54 Friday 25 September 2020 (49509)

It is possible that for for some reason also the root-filesystem NFS communication of the RTPCs has been affected, at least preventing the local login.

Again, this problem will be solved by the foreseen RTPC platform update to CentOS7 on the local disk.