VizeL Posted November 24, 2014 Share Posted November 24, 2014 Is there any significant difference in terms of ram usage on dockers vs plugins? I'm trying to diagnose why my server has had a significant slowdown, particularly when multiple streams are being attempted, or whenever a file is being moved Quote Link to comment
Pducharme Posted November 25, 2014 Share Posted November 25, 2014 I suggest using cAdvisor docker to see the resources usages of your dockers, will help diagnose. Quote Link to comment
VizeL Posted November 25, 2014 Author Share Posted November 25, 2014 thanks, will install that upon getting home! Do you know what kind of ram usage I should be expecting by chance? (I ask only because I do not have anything to compare it too -- did not keep logs of resource usage as plugin, so Im not sure if I'll see anything thats out of the norm, unless its painfully obvious) Quote Link to comment
smdion Posted November 26, 2014 Share Posted November 26, 2014 thanks, will install that upon getting home! Do you know what kind of ram usage I should be expecting by chance? (I ask only because I do not have anything to compare it too -- did not keep logs of resource usage as plugin, so Im not sure if I'll see anything thats out of the norm, unless its painfully obvious) RAM is going to be harder to track. - http://www.linuxatemyram.com/ My understanding in cAdvisor is that the "hot" is what is being used "total" includes the buffer. Quote Link to comment
jonp Posted November 26, 2014 Share Posted November 26, 2014 thanks, will install that upon getting home! Do you know what kind of ram usage I should be expecting by chance? (I ask only because I do not have anything to compare it too -- did not keep logs of resource usage as plugin, so Im not sure if I'll see anything thats out of the norm, unless its painfully obvious) RAM is going to be harder to track. - http://www.linuxatemyram.com/ My understanding in cAdvisor is that the "hot" is what is being used "total" includes the buffer. I did some testing on this a little while back (need to redo these tests with our current build sometime soon): So with plugins, whether they are running or not makes no difference as they consume space in RAM no matter what. With Docker, Containers don't start to utilize RAM until they are started, and when they are started, they don't make much of a difference. Docker itself consumes about 10MB of space in RAM as compared to running unRAID without it, so the Docker engine itself isn't a major concern. Looking at the chart, here's some things worth noting: the first row (unRAID Boot) refers to how much RAM is consumed after the OS is booted with no plugins or containers "started" but still present on the system, but with the array in a stopped state. the second row (unRAID Array Start) refers to how much RAM is consumed after the array itself is started (again, no plugins or containers have been started yet). each subsequent row is the amount of additional RAM each application consumes once started after the previous application keep in mind none of these plugins or containers were actually used during this test, this was just the basic apps themselves running in an unconfigured state RAM consumption during active use shouldn't differ between containers and plugins as they utilize RAM the same way from this point forward, except with Containers, you can actually limit the amount of memory a container has access to if you want Quote Link to comment
smdion Posted November 27, 2014 Share Posted November 27, 2014 thanks, will install that upon getting home! Do you know what kind of ram usage I should be expecting by chance? (I ask only because I do not have anything to compare it too -- did not keep logs of resource usage as plugin, so Im not sure if I'll see anything thats out of the norm, unless its painfully obvious) RAM is going to be harder to track. - http://www.linuxatemyram.com/ My understanding in cAdvisor is that the "hot" is what is being used "total" includes the buffer. I did some testing on this a little while back (need to redo these tests with our current build sometime soon): So with plugins, whether they are running or not makes no difference as they consume space in RAM no matter what. With Docker, Containers don't start to utilize RAM until they are started, and when they are started, they don't make much of a difference. Docker itself consumes about 10MB of space in RAM as compared to running unRAID without it, so the Docker engine itself isn't a major concern. Looking at the chart, here's some things worth noting: the first row (unRAID Boot) refers to how much RAM is consumed after the OS is booted with no plugins or containers "started" but still present on the system, but with the array in a stopped state. the second row (unRAID Array Start) refers to how much RAM is consumed after the array itself is started (again, no plugins or containers have been started yet). each subsequent row is the amount of additional RAM each application consumes once started after the previous application keep in mind none of these plugins or containers were actually used during this test, this was just the basic apps themselves running in an unconfigured state RAM consumption during active use shouldn't differ between containers and plugins as they utilize RAM the same way from this point forward, except with Containers, you can actually limit the amount of memory a container has access to if you want All of this is also assuming every docker being used has the same base image as well? I know the ones needo/gfjardim/smdion/hurriance use phusion. This could change dramatically if you pull in other dockers Quote Link to comment
jonp Posted November 27, 2014 Share Posted November 27, 2014 thanks, will install that upon getting home! Do you know what kind of ram usage I should be expecting by chance? (I ask only because I do not have anything to compare it too -- did not keep logs of resource usage as plugin, so Im not sure if I'll see anything thats out of the norm, unless its painfully obvious) RAM is going to be harder to track. - http://www.linuxatemyram.com/ My understanding in cAdvisor is that the "hot" is what is being used "total" includes the buffer. I did some testing on this a little while back (need to redo these tests with our current build sometime soon): So with plugins, whether they are running or not makes no difference as they consume space in RAM no matter what. With Docker, Containers don't start to utilize RAM until they are started, and when they are started, they don't make much of a difference. Docker itself consumes about 10MB of space in RAM as compared to running unRAID without it, so the Docker engine itself isn't a major concern. Looking at the chart, here's some things worth noting: the first row (unRAID Boot) refers to how much RAM is consumed after the OS is booted with no plugins or containers "started" but still present on the system, but with the array in a stopped state. the second row (unRAID Array Start) refers to how much RAM is consumed after the array itself is started (again, no plugins or containers have been started yet). each subsequent row is the amount of additional RAM each application consumes once started after the previous application keep in mind none of these plugins or containers were actually used during this test, this was just the basic apps themselves running in an unconfigured state RAM consumption during active use shouldn't differ between containers and plugins as they utilize RAM the same way from this point forward, except with Containers, you can actually limit the amount of memory a container has access to if you want All of this is also assuming every docker being used has the same base image as well? I know the ones needo/gfjardim/smdion/hurriance use phusion. This could change dramatically if you pull in other dockers No. These have several different base images in my test. Once we release an official lime tech docker base image, containers that share its usage will benefit from even less memory and disk consumption. Quote Link to comment
VizeL Posted November 27, 2014 Author Share Posted November 27, 2014 Would a dying cache drive slow down all other drives? Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.