touchup runs have resource handling leak that often crashes Solr servers (connection release?)

[kltm]

It seems that there is something that goes on in touchup that makes it not give up resources when connecting to Solr. After a while, the large number of connections in some weird open state brings the server down. Given this, until fixed, touchup should not be run against resources that cannot be easily manually restarted.

From a recent email thread:

"...trying to run (touchup) that seemed to bring down labs [GOlr dev server]... It was though[t] that production might be fine with it (Re: [go-software] High-prioty: fix production loads [was Fwd: GOJIRA amigo search issues]). ...I think is probably aiming at production now. If this is what happened, I believe there is a resource handling leak in her code that essentially causes a DDoS attack with open connections..."

"We had a lot of connections that were in some strange TCP state. Restarting nginx and jetty seems to have fixed the problem."

[kltm]

Very suspicious timing in a recent local GOlr outage seems to indicate that this is still an issue.

@selewis Also, please refrain from using the public GOlr server until this is fixed and tested--we cannot risk outage and intervention on current production systems.