Storage device control

["Alan M. Carroll" <am...@network-geographics.com>]

I am working on command line access to the state of physical storage in the cache. Currently the interface looks like

traffic_line --device /dev/sdb --cmd offline

The --device must match a path in storage.config and if so, that physical device is marked offline as if it had failed. I'm in the final stages of cleanup for this feature, so it's time to comment on it if you think it could be better.

The use case is for larger sites that have other mechanisms for monitoring disk health and want to be able to take a disk out of ATS before it causes problems (e.g., if the disk is hasn't quite completely failed it can be very slow instead of actually failing).

Implementing this exposed an existing bug, TS-2500.

P.S. I plan, someday, to add a command to bring a disk back online. So the values for --cmd should be

1) "up" / "down"
2) "online" / "offline"
3) "enable" / "disable"

I also hope to document how the command passing works, in case there are other commands that would be useful to make available via traffic_line.

Re: Storage device control

[James Peach <jp...@apache.org>]

On Jan 16, 2014, at 8:56 PM, Leif Hedstrom <zw...@apache.org> wrote:

> 
> On Jan 16, 2014, at 7:33 PM, Alan M. Carroll <am...@network-geographics.com> wrote:
> 
>> I am working on command line access to the state of physical storage in the cache. Currently the interface looks like
>> 
>> traffic_line --device /dev/sdb --cmd offline
>> 
>> The --device must match a path in storage.config and if so, that physical device is marked offline as if it had failed. I'm in the final stages of cleanup for this feature, so it's time to comment on it if you think it could be better.
> 
> 
> I don’t feel particularly strongly about it, but I probably would have preferred e.g.
> 
> 	traffic_line —offline /dev/sdb
> 	traffic_line —online /dev/sdb
> 
> 
> etc. It seems more inline with how things generally works already. I can see the original proposal getting confusing if you later add commands which should not take a —device option, but something else.  It then gets possible to postulate nonsensical combos.

I don't have a strong opinion on the command syntax, but I think that it would be useful to be able to query for status.

J

Re: Storage device control

[Leif Hedstrom <zw...@apache.org>]

On Jan 16, 2014, at 7:33 PM, Alan M. Carroll <am...@network-geographics.com> wrote:

> I am working on command line access to the state of physical storage in the cache. Currently the interface looks like
> 
> traffic_line --device /dev/sdb --cmd offline
> 
> The --device must match a path in storage.config and if so, that physical device is marked offline as if it had failed. I'm in the final stages of cleanup for this feature, so it's time to comment on it if you think it could be better.


I don’t feel particularly strongly about it, but I probably would have preferred e.g.

	traffic_line —offline /dev/sdb
	traffic_line —online /dev/sdb


etc. It seems more inline with how things generally works already. I can see the original proposal getting confusing if you later add commands which should not take a —device option, but something else.  It then gets possible to postulate nonsensical combos.

Just my $.01,

— Leif