More Spideroak pain – replacing files with error messages
Aside from the usability niggles, I did notice some more concerning behaviour that seems that it originated from Spideroak.
I opened a Powerpoint .PPTX file one day, to find that it reported the file as corrupt. Puzzled, I opened it in a text editor, and saw this:
I found another similar file – same issue. And another. And another. I backed up, and did a search for ‘Tiny’ files in Windows Explorer, and sorted by ascending size. These were all PDFs, PPTXs, and other files of at least a few hundred KB.
There were lots! In total, around 150 files had been overwritten with a server error message.
I noted this, and checked my Spideroak backup. i saw that all the files suddenly went from their original file size to 75 bytes (the error message) on the same day.
I wondered if perhaps the Spideroak sync service had suffered an outage on that day, and the sync client had erroneously synced the error message as the file! Or maybe the client couldn’t handle some local network issue, and caused that error. The fact it’s a classic server error made me think it was more likely to be the former.
OK, so, that established, I started restoring.
Crashplan is a beautiful tool here; you can specify the date and time of before whatever data loss you suffered, and it’ll let you restore whichever backup version it had saved of the affected files before then. So, I couldn’t quite drag-select and right-click the tiny files listed in WIndows explorer and restore them that way, but i could at least run through the list manually, checkboxing their counterparts in Crashplan, and on pressing ‘Go’ it would restore the relevant backup version, directly over the corrupt file on disk.
Spideroak is… not so elegant. Pretty bad, actually. On the plus side, you can right-click an original file and view in the Spideroak client, which shows the versions – so you can work directly off the ‘hit list’ of 75byte files, but only one at a time. On the downside, you can’t select ‘a point in time’ to restore from, as Crashplan can – you have to select the version of each file manually, and then restore it. Understandable for a file sync product with versioning, but not as good as a dedicated backup product.
But the worst is yet to come – although Crashplan lets you restore to the original folder, it does not overwrite the bad file! Instead, it places a copy of the restored file as a “….(1).” copy. So – after you’ve restored the 150… 250… 1000 – however many files you’ve corrupted, you then have to go through each folder manually deleting the corrupt file, and renaming the “(1)” copy to remove the “(1)”. And, of course, this is then seen as a new file by Spideroak, and uploaded to the cloud all over again!
Again, I can see this is a corner case. If the file was completely deleted, I can see that Spideroak wouldn’t place the (1) extension. But the fact they do, and Crashplan simply overwrite the file (as you asked it to do), is to highlight that they’re not a fully fledged backup product, despite claiming to be that too.
It’s a shame for me; I really liked the Spideroak concept, but now that i’m using the tool – it’s a little rough around the edges, not quite up to its peers. And the corrupted files is very concerning!