Eru0194

New Member
Jun 13, 2019
10
2
3
I recently purchased an SC846B chassis and I'm trying to find out which rails are the best for this. It seems the best options are:
This is the clearest comparison I could find. The biggest difference seems to be that MCP-290-00053-0N (yellow label) rails extend far enough to remove the lid of the chassis while it's racked. This is obviously ideal, but I'm not sure how safe it is.

The problem is that Supermicro can't/won't approve the MCP-290-00053-0N rails for their 4U chassis. I know many people are using them anyway, and I'm wondering if anyone has seen these rails fail when they're used with a fully-loaded 4U chassis. I haven't been able to find maximum load specs for any of these rails.

I've read every post I could find about SC846 rails. I'm hoping some people that have been using their 4U chassis with MCP-290-00053-0N (yellow label) rails for a long time might share their experience here.

I'm not worried about my rack tipping, just the possibility of rails failing while fully extended. Has anyone seen or heard of this happening?
 

j_h_o

Active Member
Apr 21, 2015
449
103
43
California, US
FWIW, I have an SC846 that is fully loaded with drives, and have not had an issue with the rails failing. It's been in this configuration for probably 6+ years, and I've worked on this server over 30 times, leaving the server open (while working on it) each time.
 
  • Like
Reactions: Eru0194

Eru0194

New Member
Jun 13, 2019
10
2
3
FWIW, I have an SC846 that is fully loaded with drives, and have not had an issue with the rails failing. It's been in this configuration for probably 6+ years, and I've worked on this server over 30 times, leaving the server open (while working on it) each time.
Thank you. To be clear, are you using the yellow label rails that fully extend?
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,519
964
113
artofserver.com
This is how I think about it... between the 2 full depth rack options (MCP-290-00053-0N, MCP-290-00057-0N):

If you use the "blue" label, the rails are effectively useless (without workarounds) because if the purpose of sliding out your server is to open it, you can't. So, what happens then? You have 2 workarounds:

1) you unlatch the lock, and slide it 3-4 inches further out just on the inner rail to be able to open the lid. this seems more dangerous than using the "yellow" label in my opinion. For one, you're extending it with only the inner rail, instead of in combination with the middle rail. And secondly, the rail is unlatched so it could slide all the way off. This all seems less safe.

2) you leave the 1U space above the server vacant, so you can reach into the space to open the lid. this seems the safer option, but wastes the 1U space and you can't block it off.

OR.. just use the "yellow" label rails.
 
  • Like
Reactions: nikalai

Fritz

Well-Known Member
Apr 6, 2015
2,080
457
83
66
My 2 cents worth -

I have 4 SC486 chassis's fully loaded and using the yellow label rails. I fully extend them all the time and have never had a rail failure.
However, I have had the inner rails come off the tracks on others chassis/rails and realize that if this were to happen on a fully loaded SC 846 it would coming crashing to the floor. I therefor err on the side of caution and place a lift table under the chassis when I fully extend it. It's not always high enough but it greatly reduces the distance a chassis would fall if the rails failed. :)
 
  • Like
Reactions: BLinux

BLinux

cat lover server enthusiast
Jul 7, 2016
2,519
964
113
artofserver.com
My 2 cents worth -

I have 4 SC486 chassis's fully loaded and using the yellow label rails. I fully extend them all the time and have never had a rail failure.
However, I have had the inner rails come off the tracks on others chassis/rails and realize that if this were to happen on a fully loaded SC 846 it would coming crashing to the floor. I therefor err on the side of caution and place a lift table under the chassis when I fully extend it. It's not always high enough but it greatly reduces the distance a chassis would fall if the rails failed. :)
I have 846 servers fully loaded with the yellow label rails and have never had a problem either. But, I really like your idea of putting a cart or table underneath it! That's a really nice suggestion, at least the server won't be falling on your toes.

BTW, regarding the "inner rails" coming off the tracks, I've seen that with mismatched rails between so-called "Rev A" (not labeled as such) and Rev B rails. Is that what happened? or was it a properly matched rail set that came apart?
 
  • Like
Reactions: Fritz

Fritz

Well-Known Member
Apr 6, 2015
2,080
457
83
66
I have 846 servers fully loaded with the yellow label rails and have never had a problem either. But, I really like your idea of putting a cart or table underneath it! That's a really nice suggestion, at least the server won't be falling on your toes.

BTW, regarding the "inner rails" coming off the tracks, I've seen that with mismatched rails between so-called "Rev A" (not labeled as such) and Rev B rails. Is that what happened? or was it a properly matched rail set that came apart?
Not sure if they were mismatched or not but it is a possibility. Next time I have the server in question out I'll check. Thanks.
 

ecosse

Active Member
Jul 2, 2013
374
64
28
Server rails generally are the invention of the devil. Never has so little cost so much and yet been the source of so much frustration.

Having said the, the Supermicro and HP rails I've had have generally been solid.
 

n17ikh

Member
Jul 12, 2019
41
40
18
So I have actually had a Supermicro rail fail catastrophically. It was an SC836 chassis fully loaded with drives with MCP-290-00053-0N "yellow label" rails from ebay. I extended the server to do some work and one of the rails had the inner rail and slider separate from the outer rail, dropping the server and spraying tiny ball bearings all over the place. I was able to catch it with my knees and save it from hitting the ground but it was pretty hairy for a bit, especially since all the cables on the rear were still plugged in and not long enough to deal with sitting the server on the ground!

I don't fully trust the Supermicro rails; they seem cheap and prone to that kind of failure mode. Before I had Supermicro equipment I had some older Poweredge servers (PE2950) and the Dell rapid rails for those were built a lot more solidly.

I'll have to check when I get home whether the inner and outer rails were mismatched in revision, I didn't realize those had a rev A/B. That particular chassis had the inner rails when I got it and the outer rails were purchased separately.
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,519
964
113
artofserver.com
I'll have to check when I get home whether the inner and outer rails were mismatched in revision, I didn't realize those had a rev A/B. That particular chassis had the inner rails when I got it and the outer rails were purchased separately.
glad there were no injuries from that incident!

the differences in "rev A" (which is the non-labeled version) vs "rev B" are described in this video:


in case that helps identify the combination of rails you have. please let us know if it was mismatched or if it was properly matched and still failed.
 

nikalai

Member
Oct 26, 2018
33
6
8
I searched recently for two sets of rails for two Supermicro 846 servers. In Europe are hard to find and too expensive. So i ordered MCP-290-00014-00, those are the "old style" but also much cheaper. I searched recently for two sets of rails for two Supermicro 846 servers. In Europe are hard to find and too expensive. So i ordered MCP-290-00014-00, those are the "old style" but also much cheaper.

chp.JPG

The package is in transit.
 

JSchuricht

Active Member
Apr 4, 2011
163
53
28
Be careful with those old style rails. The rollers in front fail and let the server fall. My first SC846 came with those when I bought it in 2009. I have to keep it mounted at a height I can support with my knees as both sides have failed.
 
  • Like
Reactions: nikalai

n17ikh

Member
Jul 12, 2019
41
40
18
glad there were no injuries from that incident!

the differences in "rev A" (which is the non-labeled version) vs "rev B" are described in this video:


in case that helps identify the combination of rails you have. please let us know if it was mismatched or if it was properly matched and still failed.
Looks like both inner and outer rails are rev A on the failed set. Could be that's part of why they had a new revision.
 
  • Like
Reactions: Eru0194