<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On 6 August 2014 18:32, John Meinel <span dir="ltr"><<a href="mailto:john@arbash-meinel.com" target="_blank">john@arbash-meinel.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">That sounds like what I would have expected was happening (we only run upgrade steps if we think they have a reason to change things, and then only set the last known version once all the upgrade steps have finished.)</div>
</blockquote><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>
<br></div><div>I'm concerned about the "if there is a mongo problem", because if we're running into problems for things we think are idempotent, then we might have a genuine problem with them.</div></div>
</blockquote><div><br></div><div>Sorry, I wasn't clear. I didn't mean to imply there were any problems with existing upgrade steps. I was referring to the recent mongo replicaset problems we've been having - they were sometimes preventing the upgrade-steps worker from finishing so we never came out of "upgrade mode" causing many "upgrade in progress" messages to be spewed to the logs.</div>
<div><br></div><div>By making the change I'm proposing, we can avoid the unnecessary and confusing log noise in the case where no upgrade needs to be performed.</div></div></div></div>