2.16.4: firstboot phase: hostkey exchange, syncfiles very slow (sles15.3)

[wrussian]

trafficstars

I'm using version 2.16.4 on a RH7 MN. The client image is based on sles15.3. The installation process is successful (see attached extracted messages from computes.log for time stamps and xcat.log of the client). The installation and preparation takes about 20 mins, which is ok. Anyway after the first boot (Aug 16 12:25:29) the setup of hostkeys and execution of the syncfiles takes about 37 mins. The first boot phase ends at 13:02:

Aug 16 13:02:56 xCAT_CLIENT_HN xcat.deployment INFO finished firstboot preparation, sending request to IP_MPS_MN:3002 for changing status...

Using the same MN and client for a sles12.5 installation the same 'process' takes about 5 mins. Also the xcat.log will contain the cat of the ssh - keys, which is not the case for the sles15.3:

An example is:

• getcredentials.awk ssh_rsa_hostkey
• grep -E -v '</{0,1}xcatresponse>|</{0,1}serverdone>'
• sed -e 's/</</' -e 's/>/>/' -e 's/&/&/' -e 's/&quot/"/' -e 's/'/'''/'
• grep -E '' /tmp/ssh_rsa_hostkey
• '[' 1 -ne 0 ']'
• cat /tmp/ssh_rsa_hostkey
• grep -E -v '</{0,1}errorcode>|/{0,1}data>|</{0,1}content>|</{0,1}desc>'
• logger -t xcat.deployment.postscript -p local4.info ssh_rsa_hostkey ++ cat /etc/ssh/ssh_host_rsa_key
• MYCONT=
• MAX_RETRIES=10
• RETRY=0
• '[' -z '' ']'
• '[' 0 = 0 ']'
• let SLI=32550%10
• let SLI=SLI+10
• sleep 10
• RETRY=1
• '[' 1 -eq 10 ']'

This section is repeated ten times and the download of each file listed in synclist takes about 2 min / file.

Any advice to debug this problem or solve it is highly appreciated.

Attached files: mpsInstall.log <- log entries sles15.3 from computes.log mpsxcat.log <- sles15.3 client xcat log

in both files the following replacement have been made:

xCAT_CLIENT_HN : hostname sles15.3 client IP_xCAT_CLIENT : IP sles15.3 client xCAT_MN_HN : hostname of the MPS MN IP_MPS_MN : IP address xCAT management node mps-logs.zip