Thu, 19, 2006 at 04:00:40PM -0500, Yuri Lukin wrote:
Jared Mauch wrote

What I don't get is why the "router" has to reboot twice to take
these settings into account when you resize the tcam. The 'pending reload'
that it enables isn't even visible in the "sh reload" command.
- Jared

I mentioned this in my conversation with a TAC engineer earlier today,
he denied any knowledge of this.

It does show in the "sh mls cef max" output, but not
in the regular u/i

Then eventually the box will reload with

* 17 17:56:47.517: %SYS-SP-5-RELAD: Reload requested Reload Reason: FIB Protocol Allocation mismatch.

on the console.

very lame imho, but it's a one time hit so not too bad.
- jared

Thu, 19, 2006 at 04:17:01PM -0400, Jared Mauch wrote:
Then eventually the box will reload with

* 17 17:56:47.517: %SYS-SP-5-RELAD: Reload requested Reload Reason: FIB Protocol Allocation mismatch.

on the console.

Wierd, never seen that one.

very lame imho, but it's a one time hit so not too bad.

Unfortunately that "one time hit" is way too long. Improving the boot time
should be a serious Cisco goal, and everyone who has influence should
request it. If they could have found a way to split the SP and RP images
while still achieve the native "effect", and have them both boot
simultaniously, they could have seriously reduced the load time AND
avoided the whole "images are too big for my flash" issue. Not to mention
that it seems the stock internal 64mb flashes load twice as fast (2MB/s)
as any compact flash (1MB/s), external or internal.

Stuff happens to the best of platforms, but a box crapping out is a lot
less painful when it is back up and working in under 30 secs, compared to
5+ minutes. This is still one of Foundry's biggest selling points in my
mind. :)

I don't suppose any of this has been improved for the RSP720/MSFC4?

Thu, 19, 2006 at 06:27:07PM -0400, Richard A Steenbergen wrote:
Thu, 19, 2006 at 04:17:01PM -0400, Jared Mauch wrote:
Then eventually the box will reload with

* 17 17:56:47.517: %SYS-SP-5-RELAD: Reload requested Reload Reason: FIB Protocol Allocation mismatch.

on the console.

Wierd, never seen that one.

very lame imho, but it's a one time hit so not too bad.

Unfortunately that "one time hit" is way too long. Improving the boot time
should be a serious Cisco goal, and everyone who has influence should
request it. If they could have found a way to split the SP and RP images
while still achieve the native "effect", and have them both boot
simultaniously, they could have seriously reduced the load time AND
avoided the whole "images are too big for my flash" issue. Not to mention
that it seems the stock internal 64mb flashes load twice as fast (2MB/s)
as any compact flash (1MB/s), external or internal.

Stuff happens to the best of platforms, but a box crapping out is a lot
less painful when it is back up and working in under 30 secs, compared to
5+ minutes. This is still one of Foundry's biggest selling points in my
mind. :)

I don't suppose any of this has been improved for the RSP720/MSFC4?

The modular software is a big win for this. I would
suggest refusing to purchase any platforms from cisco that
don't include this fancy feature called "protected memory".

The legacy issues are getting to be too much of a liability IMH

I've tried to explain the whole boot time issue to folks they
seem to not care that foundry is wiping the floor with them. All I
hear is "redundant processors" which dosen't solve the true issue. They've
all missed the mark (many vendors).
- Jared