Rescue and log Net::ReadTimeout when sending oa email notifications

[ajkiessl]

In OpenAccessNotifier#send_notification_to_user, we rescue Net::SMTPFatalError and log it so the iterator continues to the next notification. We have, however, run into other Net errors. Specifically: ReadTimeout. This could be caused by a temporary issue with the connection to the mail server, so I think it makes sense to rescue it as well, log it, and move to the next notification.

[ajkiessl]

We may also want to consider moving the time when these emails go out. Seems to be more common that there are network issues Sunday night. We had these emails going out Monday mornings before infrastructure changes.

[anaelizabethenriquez]

related to #959 (timing issue)

[ajkiessl]

We also might want to look in to how large the email is that fails. It could be that there's a user with a lot of publications listed, and there's an email that times out because it takes a while to send.

[anaelizabethenriquez]

Interesting. If so, that would be another reason (besides usability) to address #735.

[anaelizabethenriquez]

In case anecdotal evidence is helpful, I checked the date on a recent reminder email sent to a very prolific researcher (4279). They typically get very long emails, so I wondered if it would have led to a time-out. However, their latest email went out November 3, when a full batch sent. Since they replied to it, I can tell you that it mentioned 8 new publications and 32 previous ones. Maybe something other than the size of the email is causing the time-out?

[ajkiessl]

• Rescue Net::ReadTimeout when sending OA notifications. #1028
• Investigate what is causing the Net::ReadTimeout

[ajkiessl]

I ran the notification task in QA and it did not raise any errors. Waiting to see if the rescue code catches anything in production.