GitLab.com outage on 2015-09-01

This post details a specific incident of GitLab.com being offline for an hour due to perceived filesystem corruption on an NFS server. It describes the initial misdiagnosis based on `dmesg` output, the decision to take the site offline for `fsck`, and the subsequent realization that the errors were recurring daily and not the cause of the immediate slowdown. The post highlights the challenges of diagnosing NFS issues, the decision to abort `fsck` and bring the site back online, and the ongoing effort to migrate data off the problematic filesystem. It also reflects on lessons learned regarding communication during crises and the need for better status updates.