Amiga crash protection

I was wondering how the Amiga crash detection routines work ?
In what order ?

A was thinking whether Amiga could have a way to detect a crash and swap RAM: dynamic RAD: or any file being worked on to RAD: as a safeguard.

I know developers are supposed to work in RAD: to recover their files in case of a crash :)

@angelheart
"work in RAD: to recover their files in case of a crash"

Except on x1000

@angelheart

Well you put stuff on RAM: because you want it to be temporary.

RAD: is different is work more like floppy disk, if you have ADF image you can RAW write to the RAD device, reboot your Amiga whit warm boot, and boot from RAD. Warm reboot and the RAD is empty.

Developers do not want there work to be lost, they don't use RAD, they use Raid, Raid has nothing to do whit RAD, Raid is tenontology for recrudency.

Basically two hard drives are mirroring etch other, so if one HD breaks, no work is lost, because it also exist on a other HD.

Another thing developers use is GITHUB and SourceForge, for uploading changes to the source code. So if the house burns down, at least you have your source code some where else.

If you want to find and recover deleted files you, need a SFS partition, you can find deleted files in .rescyled directory its hidden.

On FFS partitions you need to use the recovery tool.

@angelheart

Quote:

By a exception handler, there is no order, the exception handler will trigger when some thing illegal is detected.

The MMU or CPU, will notice some thing wrong and trigger a interrupt, this triggers a interrupt routine, (exception handler)

Writing to Memory that is Write protected (DSI error.)
Reading form memory that is illegal to read from (DSI error)
Corrupted program code, or program code not found (ISI error)

Then “Grim Reaper” is started to inform the user or developer what happened.

Some crashes can trigger the older Guru Mediation, this normally triggered by the some Guru messages routines, this happens when Libraries or devices, or other core part of the OS notice some thing bad, can also be triggered by older programs.

@LiveForIt
Quote:
First the trap handler tries to handle the issue (it dumps some stuff out the serial port). Then the Reaper, built into the kernel, is activated (dumping stuff out the serial port). Then the Reaper will try to launch GrimReaper.

@all
Here is a totally awesome tutorial (I'm slightly biased):
http://wiki.amigaos.net/index.php/AmiWest_2013_Lesson_1

Once you have a crash report, you'll find this useful:
http://wiki.amigaos.net/index.php/AmiWest_2013_Lesson_2

@LiveForIt

Thanks interesting, anything for power loss detection or delay to save files ?

I was just thinking about possibility to save anything recoverable to RAD: (not RAID) in rare cases or even file corruption detection. Maybe even a RAD: like device for the system to use in such event.

@ssolie
I forgot about serial !

Hmm so a serial dump redirect to a SystemRad: could be possible ? - just saw on the page ssolie gave - DumpDebugBuffer survives a boot.

Print queue in progress on power outage, delayed redirect to SystemRad: ?
Other ways to use such technology ?

Thanks @all

@angelheart

Quote:

Only a soft reboot. If your crash causes or requires a hard reset to get over, then serial is the only option for capturing the log.

Quote:

Power outage implies a the hardest of hard resets, Rad: won't survive that! If you want a printer spool that survives power reset it'll need to be hard disk based.

@broadblues

Quote:

Yes, I was looking into crash protection and power outage protection, in an attempt to see if under such an event a 'delay in shutdown' could possibly be used, possibly by RAD: to attempt emergency auto save of work/ prefs/ transmissions.

I was thinking also into hibernation, instant on, restart system state which has windows, programs, tabs exactly as they were when a shutdown unexpected.

Disk based or cloud based incremental updating etc.

even gameplay freeze states etc.

what does it take for AmigaOS to handle all this ?

What are the official libraries / frameworks to access online servers / dbs etc ?

@angelheart

Quote:

in a case where need to keep your PC safe, you can get a small UPS, the sate of UPS can be checked by querying SNMP.

possibly by RAD: to attempt emergency auto save of work/ prefs/ transmissions.

It does not make sense when you need to power anything in RAM will lost, and RAD in a RAM drive so anything there will be lost.

Quote:

It used to be possible to save the state of a Amiga500 using a Action Replay III. How ever its a bad thing to save a state of an computer in middle of a write operation, this will need to be handled whit a lot of care.

Quote:

A unexpected crash will result in system that may be unsafe, and can result in other crashes, restoring a crashed system will bad idea.

A unexpected shutdown, When a some program ask the computer shutdown instead of user. In this case saving work can be handled by Arexx scripts and Close Window Event to close open programs safely, before the shutdown down.

Sending a CTRL+C to programs it not a smart way stop programs, it can often result in shell programs stopping unexpected before closing files, libraries and devices, and so it might result in a lot system wide problems.

In other words there are no 100% safe ways stop programs, I think it vice to time to time reboot to get a fresh system.

@angelheart

Quote:

BsdScoket.Library for low level UDP/TCP stuff.

Quote:

HTML / FTP / and other service protocols there are lots, this are not build into the OS, you can find programs or libraries some one else has written (mostly for Linux), some thing like LibCurl maybe.

Many things are shell command based like SVN and GIT, not some thing you can access from a framework, only whit a lot tricks.

Quote:

Maybe ODBC?
Some web based databases can be handled by HTML.

Dare I ask why not just save important files to your harddisk...? Works for me