Troubleshooting
Problem
You are transferring exports using SCP (e.g. audit results, datamarts) to a Linux host and they fail. There are errors in the guard_filetransfer_log similar to:
15965 Thu Aug 29 06:30:00 2019 guard_filetransfer.pl: Starting file transfer.
15965 Thu Aug 29 06:30:00 2019 scp STDOUT> spawn /usr/bin/scp -q -P 22 /var/tmp/MyReport.csv.gz guarduser@host.com:/mypath/reports/
15965 Thu Aug 29 06:30:01 2019 scp STDOUT> lost connection
15965 Thu Aug 29 06:30:01 2019 scp STDOUT> RESULT=connection refused
15965 Thu Aug 29 06:30:01 2019 SCP to: host.com, User: guarduser, Path: /mypath/reports/, File: /var/tmp/MyReport.csv.gz
15965 Thu Aug 29 06:30:01 2019 Error code: 1
15965 Thu Aug 29 06:30:01 2019 transfer to primary and secondary host failed or there is only primary and transfer to it failed
Many exports are scheduled around the same time from multiple appliances to the same remote host.
Cause
The remote host is configured to receive a maximum amount of connections that sshd can handle. This limit is too low for the number of appliances that are concurrently sending data.
Diagnosing The Problem
With verbose logging for the SCP command, this error is seen right before the "lost connection":
scp STDOUT> ssh_exchange_identification: Connection closed by remote host
Retrying the export works. You can see in the guard_filetransfer_log that an export previously failed, was retried, and then it succeeded.
Run this on the remote server to see how many connections sshd can handle:
sudo grep MaxStartups /etc/ssh/sshd_config
Resolving The Problem
There are two ways to resolve this:
1. Reschedule the exports staggering the times.
2. Increase the MaxStartups parameter on the remote host.
Document Location
Worldwide
Was this topic helpful?
Document Information
Modified date:
11 November 2019
UID
ibm11079091