Notices

Is there a way to prevent the SD card from unmounting when the back comes off. I have had several pocket drops where I loose the battery door but the battery stays in, it still kills the system and about 30% requires backupmenu and a good backup.

While completely unrelated to FlopSwap the easy way to do this is glue a small powerful magnet over the back cover sensor. This tricks the device into thinking the back cover is closed and stopping swap from unmounting. There is a thread somewhere about this.

In my one or two unfortunate cover removals while live, I have not had any trouble after a reboot. Not to say corruption couldn't happen.

Just to close the loop on this, I decided to RMA my SanDisk 64gb card. The replacement that SanDisk sent me works perfectly. I've been using swap on the card without any random reboots for a week or so now.

I am seeing "0% fresh swap written" every time I Check Status.

The only possible culprit is that I've just installed Easy Debian while swap was on the SD card (only). Easy Debian, I think, wanted to restart the swap but tried to do it on the device swap, and then reported something like a wrong argument (/dev/mmcblk0p0) for swapoff. I checked /proc/swaps and saw that swap was now both on the device and on the SDcard. I rebooted, and told Flopswap to swap to Fresh Swap. It is showing (Show Current) the correct swap file (on the SD card), but even after starting half a dozen programs it still says 0% fresh swap written. Please advise.

I've used Flopswap for months and have never seen this 0% business.

It is possible to get 0% Fresh Swap Written but not for long if you are using the device.

More likely the issue is that Easy Debian has moved the partitions so FlopSwap is no longer monitoring the correct partition. Because FlopSwap's partitions are "hardcoded" this would make it incompatible.

I have never used Easy Debian but I do believe it modified the partitions somehow.

ED does not modify partitions on its own. In fact, the default installation runs from an image in /media/mmc1. It may, however, be installed in a partition (most common use case is on the SD), which requires a manual preparation. Might it be possible that chill moved partitions around to fit ED and the new arrangement fooled FlopSwap?

Caveat: I have never used FlopSwap and cannot test.

I didn't manually move the partitions. I did have Flopswap on prior to installing ED, and swap was on the SD card. ED installation then first reported an invalid argument to swapoff (I think the argument was /dev/mmcblk0p3). Then, while the ED image file was decompressing, I did a cat /proc/swaps and saw that both the N900 and the SD card were used for swapping (the N900's swap had 0 bytes used, though). Not that this explains everything...When I installed ED before (circa a year ago), I was not swapping to SD card, and subsequent Flopswap and swapping to SD card worked fine (I'd write 94% fresh swap in just a few hours or so).

I also did cat /proc/diskstats now. Looking at Flopswap's swapused.sh, it seems that swapused.sh calculates the fresh swap correctly (and I don't see any errors in the script) - the problem is that the number of written blocks (in diskstats, the 7th argument after the device name) is too low, generally 10000-30000, and rising very slowly (so it takes hours just to get to 5% fresh swap); meanwhile Conky reports hundreds of MB of swap written.

So what could be happening to my /proc/diskstats, and how could ED cause it? Obviously this doesn't look like a Flopswap issue (anymore), but if I moved this to the ED thread I might come across as a cross-poster.

This is an issue with Easy Debian package IMO.

I have looked through the sources, Easy Debian package assumes your swap is on the device (0p3) when running the installer.

"/home/user/debian-installer.sh" - Line 145
"/sbin/swapoffon"
IMPORTANT NOTE:-
It seems anyone wishing to install Easy Debian with Swap on uSD should move Swap back to the device (0p3) before installing.

As for a fix I would hope a reboot should put it straight but you say you have tried this. I have not spotted anything in the source which would change any partitions.

The calculations are designed to increase slowly to avoid hitting "int32" (see earlier posts) However, it could be that your swap has been stripped between the card and device somehow.

Ok, my /proc/swaps looks ok:

Filename Type Size Used Priority
/dev/mmcblk1p2 partition 683000 85804 -1

but look at my /proc/partitions:

major minor #blocks name

7 0 3145728 loop0
179 0 15558144 mmcblk0
179 1 14191600 mmcblk0p1
179 2 683008 mmcblk0p2
179 3 683008 mmcblk0p3
179 8 31264768 mmcblk1
179 9 28315648 mmcblk1p1
179 10 2097152 mmcblk1p2
179 11 786432 mmcblk1p3

So my 16GB SD card is now on mmcblk0? And I am swapping to mmcblk1p2? How'd that happen? Where is the swap actually?

Meanwhile Conky correctly reports the intended (SD card) swap size, i.e. 667 MB. Fresh Swap on Flopswap also goes through.

Hmm, do you still have KP installed?
Have you installed any other updates recently?
Seems something has changed your boot setup,


--Edit

Also noticed that easy debian deletes and increases the size of the /tmp directory. This will mess up calculation of swap until a reboot.

The word "kludge" is used many times in the source of easy-debian