After moving cardfinder to heroku, there's a failure rate on cardfinder calls—newsbot headline generation seems to fail roughly every other day due to a socket hangup error.
This doesn't seem to be a persistent issue—I don't think I've seen it fail twice in a row, for instance—so it's possible that adding a simple retry on cardfinder calls would mitigate the problem.
Example failure log:
[cardSearchCustomQuery.s (post space-replace)]: t!=basic
[cardSearchCustomQuery.terms]: t!=basic
parsed query argument: not t:basic
Searching for 1 cards.
REQUEST DATA
undefined
Error: Failed to fetch card data: Error: socket hang up
at cardFinderSearch (/var/git/mtgnewsbot/src/lib/plugins/tracery/modifiers/cardfinder-modifiers.js:363:13)
at <anonymous>
at process._tickCallback (internal/process/next_tick.js:188:7)
After moving cardfinder to heroku, there's a failure rate on cardfinder calls—newsbot headline generation seems to fail roughly every other day due to a socket hangup error.
This doesn't seem to be a persistent issue—I don't think I've seen it fail twice in a row, for instance—so it's possible that adding a simple retry on cardfinder calls would mitigate the problem.
Example failure log: