-
Notifications
You must be signed in to change notification settings - Fork 43
error on ENOENT lstat could use additional context #21
Comments
Are you able to repro this with some consistency? We get it about once every 5 installs on our CI slave, always with the same from from the same package, but we're not sure what's required to have it happen every time. We also have it happen on our local dev boxes, but not as frequently. The error is always from same file. |
@dylang I've literally never seen this. (@othiym23 have you?) Do you have a package I can try to install where you see this occur? What is the same file that this happens to? This stat failure, if an ENOENT, would mean that the Writer finished writing its thing, but then after creating the thing, it didn't exist. If you're writing a symlink, and in follow:true mode (or on an OS without |
I don't have a package you can try, it's our closed-source (for now at least) build system. The file that seems the problem is Here's last part of a log. I don't want to share the entire log because I'd need to make sure it was all safe to share on the interwebs and it's really long.
My teammates might have more to share. Thanks for your attention to this, it's been painful having builds frequently fail for seemingly unknown reasons. |
@isaacs @othiym23 this manifests itself in the ticket reported here: npm/npm#5841 where I attached the full |
@isaacs I haven't seen this happen on my machine since we released 1.4.15 with the fixes around locking. Given the VAST number of issues we see around this, though, I completely believe that there's something fishy going on here. |
When a caller hits this stat failure, it takes considerable sleuthing to figure out why the stat was even happening. Wrapping the error message up with some explanatory text would be very helpful here.
The text was updated successfully, but these errors were encountered: