January 27, 2017 at 17:49How is storage different from chicken?
This is a story from the Service Engineer's Notes series. We will answer the question a little later, and first a small tale about one MASSIVE fall. Recently, a client contacted our Service Center, who decided to upgrade the HDS AMS2000 array on their own: replace some of the old disks with larger ones. According to the customer, he inserted a new disk - and one controller broke; pulled out a new disk - and the second controller also turned on the emergency indication, and access to the array was lost from all systems.
We ask the administrator questions about his actions. It turns out that it was this administrator who took up his shift after they decided to pull out a new disk. As a result, the array did not turn on. According to the logs from the controllers, the controllers could not spin up the disks, and there was no indication on the disks.
What could have happened here? We gathered a “consultation” of our service engineers, and began to think. First version: maybe something happened to the disk backplane when the disk was installed and it became inoperative? We decided to find the chassis of the array without controllers in order to replace the entire chassis together with the backplane. Not finding it, they asked the customer to transfer all the disks in the chassis with the controller and try to start the array.
And here we were waiting for what was called “hit - and catharsis” (and the very “banana skin” that dropped the whole system). It turned out that in the chassis with the controller the client pulled out exactly the first five disks (if you know what I mean)! It was their size that he wanted to increase. All the time while attempts were made to reanimate the array, the disks were removed. Returned the previous wheels in place. And, lo and behold, the array turned on!
The moral of this fable is simple: only a chicken can run headless, and even then not for long. In the HDS Mid-Range, the first five discs store a copy of the array configuration. When loading, the array first tries to spin the first 5 disks in the chassis with the controller. In case of failure (for example, due to the lack thereof), the download stops.
Nikolay Vedyashkin, expert at the Jet Infosystems Service Center.
We ask the administrator questions about his actions. It turns out that it was this administrator who took up his shift after they decided to pull out a new disk. As a result, the array did not turn on. According to the logs from the controllers, the controllers could not spin up the disks, and there was no indication on the disks.
What could have happened here? We gathered a “consultation” of our service engineers, and began to think. First version: maybe something happened to the disk backplane when the disk was installed and it became inoperative? We decided to find the chassis of the array without controllers in order to replace the entire chassis together with the backplane. Not finding it, they asked the customer to transfer all the disks in the chassis with the controller and try to start the array.
And here we were waiting for what was called “hit - and catharsis” (and the very “banana skin” that dropped the whole system). It turned out that in the chassis with the controller the client pulled out exactly the first five disks (if you know what I mean)! It was their size that he wanted to increase. All the time while attempts were made to reanimate the array, the disks were removed. Returned the previous wheels in place. And, lo and behold, the array turned on!
The moral of this fable is simple: only a chicken can run headless, and even then not for long. In the HDS Mid-Range, the first five discs store a copy of the array configuration. When loading, the array first tries to spin the first 5 disks in the chassis with the controller. In case of failure (for example, due to the lack thereof), the download stops.
Nikolay Vedyashkin, expert at the Jet Infosystems Service Center.