If system disk is SSD and backups are large, Duplicati's default options will cause write-wear

[kenkendk]

From rickyorb...@gmail.com on June 13, 2014 06:27:00

As per the title, it may be wise to prompt a first-time user to choose their cache location or provide alternative options.

Original issue: http://code.google.com/p/duplicati/issues/detail?id=1005

--- Want to back this issue? **[Post a bounty on it!](https://www.bountysource.com/issues/3562949-if-system-disk-is-ssd-and-backups-are-large-duplicati-s-default-options-will-cause-write-wear?utm_campaign=plugin&utm_content=tracker%2F4870652&utm_medium=issues&utm_source=github)** We accept bounties via [Bountysource](https://www.bountysource.com/?utm_campaign=plugin&utm_content=tracker%2F4870652&utm_medium=issues&utm_source=github).

[kenkendk]

From nickjor...@gmail.com on July 04, 2014 14:32:23

I would certainly want this feature! Altough it might not be a huge problem it can definitely become one!

[kenkendk]

From rst...@gmail.com on July 05, 2014 00:29:43

--tempdir=something in options should do what you want. Does it solve your issue?

Labels: -Type-Defect

[kenkendk]

From rickyorb...@gmail.com on July 05, 2014 11:31:40

The issue is that default options cause the problem and people will be unaware that Duplicati uses the system disk for temp files. I can change the temp dir through the GUI and command line, but that's besides the point because it took me a while before I realised it was even using my SSD.

You need to prompt the user to select their temp dir on first-run, or try to detect if the default temp storage disk is an SSD and if so, then prompt them to change it.

[kenkendk]

From rst...@gmail.com on July 06, 2014 03:38:24

I have seen wear levelling tests with current SSDs. The testers constantly wrote data to the SSDs and some devices could stand more than 1PB (1,000TB, 1,000,000GB) of data. For an estimated life time of 5 years this is about 400GB of data per day.

[kenkendk]

From rickyorb...@gmail.com on July 06, 2014 08:06:05

I have also seen some SSDs die after a few months because they were used as a system disk on a computer with not much memory. They didn't turn off the swap file or move it and the writes killed the disk.

[kenkendk]

From timo...@googlemail.com on July 06, 2014 08:33:17

I have a SSD which is used as system-disc (as its the only disc in the notebook). I am sorry: The comment regarding "tempdir" doesn't help me at all, because i haven't got any idea what "shomething" should be. Is it just the fact that i have to name "any" different directory? And i am wondering if it has to do something with issue 1004 , which i reported. thanks for any hints.

[kenkendk]

From rickyorb...@gmail.com on July 06, 2014 08:48:08

This refers to unnecessary writes on SSDs where a system can write to a non-SSD location. If all you have is an SSD, you can reduce temp writes by disabling the upload asynchronously option.

[yasinCelik1]

Is this issue still open?
I only have a SSD as system disk. Is there no option to not use a tempdir on disk? RAM memory is nowadays sufficient enough to hold a couple of hundreds of MB's...

[kenkendk]

Yes, the issue is still open. You can create a memory mapped filesystem/drive and set that as the temporary folder.

SSD drives have gotten much better at dealing with write-tear in the meantime though.

[kaovilai]

Yes, the issue is still open. You can create a memory mapped filesystem/drive and set that as the temporary folder.

SSD drives have gotten much better at dealing with write-tear in the meantime though.

I would love to see this resolved. No matter how good SSD drives are dealing with write tear via chip leveling etc it is a write nonetheless.
On windows I don't think there's a native way to map a ramdisk without external application?

[rcmaniac25]

It would probably be good to get this closed. I went to look for a Windows ramdisk and apparently if you use the native CreateFile function along with FILE_ATTRIBUTE_TEMPORARY and FILE_FLAG_DELETE_ON_CLOSE it effectively acts like a ramdisk by only writing the file "cache". I haven't looked passed the API docs, and there is still a bit of a question mark on my mind about it, but could be possible to do from the application perspective (at least on Windows).