ctsit/research-subject-mapper

Using a site-catalog.xml in GSMI with greater than 5 sites causes connection issues

Opened this issue · 0 comments

Site-catalog.xml is used to list connection details for all the SFTP sites that GSMI must connect to in order to place its output files. For small numbers of sites (<5), GSMI does not encounter any issues. When the site-catalog.xml grows to include more sites however (current UF production has 33 sites listed), it can cause GSMI to encounter connection issues.

The error messages emailed from GSMI include:
There was an error sending file smi.xml to sftp.ahc.ufl.edu: Unable to open channel.
There was an error sending file smi.xml to sftp.ahc.ufl.edu: Error reading SSH protocol banner

Because UF production is connecting to the same SFTP server consecutively, with different username/passwords each time, the issue may be caused by overloading the server with rapid consecutive connections. Sleeping for a short period of time in between connections may be helpful. See: http://stackoverflow.com/questions/25609153/paramiko-error-reading-ssh-protocol-banner, which mentions the same paramiko module that GSMI uses.