You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@maven.apache.org by "Haszlakiewicz, Eric" <EH...@transunion.com> on 2010/11/02 18:20:43 UTC
RE: Request to re-open MNG-3472
>-----Original Message-----
>From: Martin Gainty [mailto:mgainty@hotmail.com]
>Sent: Friday, October 29, 2010 1:14 PM
>To: users@maven.apache.org
>Subject: RE: Request to re-open MNG-3472
>
>Hi Brian
>
>if diskspace was a concern i would concur but as each repository and their
>mirrors support terabyte capacity i would hold off on the purge
>i was bitten by a wizbang version maven plugin that wasnt tested with
>wizbank injector and failed when the maven-plugin's were injected using
>incorrect role and roleHint
What in the world are you talking about? How can you claim that everyone's has the capability and the money to buy terabytes of disk space, especially for everyone's local repositories?
>> From: Brian.Levine@nokia.com
>> To: users@maven.apache.org
>> Date: Fri, 29 Oct 2010 15:28:27 +0200
>> Subject: RE: Request to re-open MNG-3472
>>
>> I'm not suggesting that Maven periodically run a task to purge the repo.
>I'm suggesting that a settable policy be implemented such that a max number
>of versions of a snapshot artifact be kept in the local repo. When that
>limit is exceeded, the oldest version is deleted.
>>
>> Yes, it's easy enough to do an rm -rf on your local repo. But that
>assumes that you understand why you need to do this or have been told by
>your dev team to do this and that you remember to do it before you realize
>that a lot of disk space is being used up by something.
This is even more of a problem when you have dozens of people performing builds on a variety of shared machines. It seems like many people don't realize that not everyone limits their use of maven to personal PCs with huge disks. When you have several people working on the same system "oh, it's just a few GB" quickly ends up turning into "out of disk space".
>> IMHO, a tool should never have unbounded access to resources such as disk
>space. Further, I don't think stakeholders (not all of which are
>developers) should need to understand the inner working of Maven or what
>they need to do work around a limitation such as this.
>>
>> -brian
I definitely agree with this. It's hard enough to get people to clean up the files that the explicitly create.
eric
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
Re: Request to re-open MNG-3472
Posted by Ron Wheeler <rw...@artifact-software.com>.
On 03/11/2010 1:17 PM, Haszlakiewicz, Eric wrote:
>> -----Original Message-----
>> From: Ron Wheeler [mailto:rwheeler@artifact-software.com]
>>
>> On 02/11/2010 3:29 PM, Haszlakiewicz, Eric wrote:
>>>> -----Original Message-----
>>>> From: Ron Wheeler [mailto:rwheeler@artifact-software.com]
>>>> The guys with small disk can just delete their entire local repo and let
>>>> maven rebuild it by itself from your central server which should have
>>>> lots of space.
>>>> One or two builds usually gets us back to a fast build from localhost.
>>> To delete the repo of the guy the next desk over I need to wait for him
>> to be in the office, then email or call him, then explain what I want him
>> to delete. And that's all only if I actually notice that that's what's
>> taking up all the space.
>>> Even if I'm just cleaning up my own files, why should I have to spend all
>> this time thinking about it and doing things to fix it?
>>> For example, I don't have to think about clearing out the cache in my
>> browser, why should maven's cache be any different?
>>
>> Each of my guys is responsible for his own workstation. They know what
>> maven is doing.
> Gee, it must be nice to work with a whole group of maven geniuses. The people I work with are busy writing code and their time is better spent on that than having to remembering to perform basic maintenance on the build tool.
> As for workstations, I don't care about what people do on their individual PCs. That's not where the problem is.
>
>>>> If your central server is short of space spend $100 and add a terabyte
>> or
>>>> 2.
>>> Oh, are you going to fund the additional disk space, and the extra backup
>> resources needed, and pay for time it takes to actually update the disk
>> array and provision the space in my company's SAN? If so, great!
>>> My point it that using more disk space has more costs than just the price
>> of a drive platter.
>> You are right.
>>
>> I am not sure that a Maven Repo really belongs in a corporate datacentre
>> since it contains nothing that can not be replaced and is mostly stuff
>> that comes from outside sources anyway.
>>
>> You could always move your repo to the cloud and pay a few dollars per
>> month for the storage or just add a local desktop server to the
>> development group.
> What are you talking about? I thought this conversation was about the files that maven stores in the ~/.m2 directory. How could that possibly NOT be in the "corporate datacenter" if that's where everyone is doing development (and release builds)?
> FWIW, no I *can't* just add a local desktop server. For one thing it wouldn't do any good b/c not everyone would be able to get to it (we're not all sitting in the same room, nor even the same country!). Furthermore, that would be a huge headache since it would be outside of all of the procedures we've set up to have consistent environments, periodic backups, etc...
>
~/.m2 directories are disposable and do not need to be anywhere special
nor backed up.
Your Nexus can be mirrored if you work in different locations and want
to offer LAN speeds rather than Internet speeds for developers.
We keep ours in our datacentre off-site but have thought about moving it
to the development site to get the speed.
I would not mind moving it to the cloud to get more bandwidth and less
concern about disk space.
I do not worry about the disk space particularly but I am not deploying
a lot of Snapshots to Nexus since we have a small team and higher QC
standards about what gets deployed.
Losing the Nexus database is not my biggest worry since most of the
stuff is third party and I do have the sources in Subversion to rebuild
any version of what is deployed to Nexus. I am not saying that I would
enjoy the process but it would only slow us down for part of a day if we
had to restore from a backup that was a month old. A bit longer if we
had to manually rebuild from scratch to get the 3rd party libraries that
have to be manually uploaded and to redeploy the current versions under
development from our sources.
Most of the versions of our stuff could be lost without any crying. No
one wants to put our production applications back to an earlier version
so we would only have to rebuild new stuff or something from the current
version in production.
My Subversion repo does get backed up and that would worry me if I lost
it! Nexus - not so much.
> eric
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
> For additional commands, e-mail: users-help@maven.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
RE: Request to re-open MNG-3472
Posted by "Haszlakiewicz, Eric" <EH...@transunion.com>.
>-----Original Message-----
>From: Ron Wheeler [mailto:rwheeler@artifact-software.com]
>
>On 02/11/2010 3:29 PM, Haszlakiewicz, Eric wrote:
>>> -----Original Message-----
>>> From: Ron Wheeler [mailto:rwheeler@artifact-software.com]
>>> The guys with small disk can just delete their entire local repo and let
>>> maven rebuild it by itself from your central server which should have
>>> lots of space.
>>> One or two builds usually gets us back to a fast build from localhost.
>> To delete the repo of the guy the next desk over I need to wait for him
>to be in the office, then email or call him, then explain what I want him
>to delete. And that's all only if I actually notice that that's what's
>taking up all the space.
>> Even if I'm just cleaning up my own files, why should I have to spend all
>this time thinking about it and doing things to fix it?
>> For example, I don't have to think about clearing out the cache in my
>browser, why should maven's cache be any different?
>
>Each of my guys is responsible for his own workstation. They know what
>maven is doing.
Gee, it must be nice to work with a whole group of maven geniuses. The people I work with are busy writing code and their time is better spent on that than having to remembering to perform basic maintenance on the build tool.
As for workstations, I don't care about what people do on their individual PCs. That's not where the problem is.
>>> If your central server is short of space spend $100 and add a terabyte
>or
>>> 2.
>> Oh, are you going to fund the additional disk space, and the extra backup
>resources needed, and pay for time it takes to actually update the disk
>array and provision the space in my company's SAN? If so, great!
>> My point it that using more disk space has more costs than just the price
>of a drive platter.
>You are right.
>
>I am not sure that a Maven Repo really belongs in a corporate datacentre
>since it contains nothing that can not be replaced and is mostly stuff
>that comes from outside sources anyway.
>
> You could always move your repo to the cloud and pay a few dollars per
>month for the storage or just add a local desktop server to the
>development group.
What are you talking about? I thought this conversation was about the files that maven stores in the ~/.m2 directory. How could that possibly NOT be in the "corporate datacenter" if that's where everyone is doing development (and release builds)?
FWIW, no I *can't* just add a local desktop server. For one thing it wouldn't do any good b/c not everyone would be able to get to it (we're not all sitting in the same room, nor even the same country!). Furthermore, that would be a huge headache since it would be outside of all of the procedures we've set up to have consistent environments, periodic backups, etc...
eric
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
Re: Request to re-open MNG-3472
Posted by Ron Wheeler <rw...@artifact-software.com>.
On 02/11/2010 3:29 PM, Haszlakiewicz, Eric wrote:
>> -----Original Message-----
>> From: Ron Wheeler [mailto:rwheeler@artifact-software.com]
>> The guys with small disk can just delete their entire local repo and let
>> maven rebuild it by itself from your central server which should have
>> lots of space.
>> One or two builds usually gets us back to a fast build from localhost.
> To delete the repo of the guy the next desk over I need to wait for him to be in the office, then email or call him, then explain what I want him to delete. And that's all only if I actually notice that that's what's taking up all the space.
> Even if I'm just cleaning up my own files, why should I have to spend all this time thinking about it and doing things to fix it?
> For example, I don't have to think about clearing out the cache in my browser, why should maven's cache be any different?
>
Each of my guys is responsible for his own workstation. They know what
maven is doing.
Maven would have a hard time knowing what to throw out but I suppose
that if it supported a simple rule-set (max space allowed, delete oldest
Snapshots if space needed) and the ability to turn automated cleaning on
and off it might be useful.
>> If your central server is short of space spend $100 and add a terabyte or
>> 2.
> Oh, are you going to fund the additional disk space, and the extra backup resources needed, and pay for time it takes to actually update the disk array and provision the space in my company's SAN? If so, great!
> My point it that using more disk space has more costs than just the price of a drive platter.
You are right.
I am not sure that a Maven Repo really belongs in a corporate datacentre
since it contains nothing that can not be replaced and is mostly stuff
that comes from outside sources anyway.
You could always move your repo to the cloud and pay a few dollars per
month for the storage or just add a local desktop server to the
development group.
> eric
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
> For additional commands, e-mail: users-help@maven.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
RE: Request to re-open MNG-3472
Posted by "Thiessen, Todd (Todd)" <tt...@avaya.com>.
> My point it that using more disk space has more costs than just the price
> of a drive platter.
>
It also has a lot of savings. And besides, you don't need all the frills. What you doing right now sounds very very costly. A simply investment in a bare bones server will save you a lot.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
RE: Request to re-open MNG-3472
Posted by "Haszlakiewicz, Eric" <EH...@transunion.com>.
>-----Original Message-----
>From: Ron Wheeler [mailto:rwheeler@artifact-software.com]
>The guys with small disk can just delete their entire local repo and let
>maven rebuild it by itself from your central server which should have
>lots of space.
>One or two builds usually gets us back to a fast build from localhost.
To delete the repo of the guy the next desk over I need to wait for him to be in the office, then email or call him, then explain what I want him to delete. And that's all only if I actually notice that that's what's taking up all the space.
Even if I'm just cleaning up my own files, why should I have to spend all this time thinking about it and doing things to fix it?
For example, I don't have to think about clearing out the cache in my browser, why should maven's cache be any different?
>If your central server is short of space spend $100 and add a terabyte or
>2.
Oh, are you going to fund the additional disk space, and the extra backup resources needed, and pay for time it takes to actually update the disk array and provision the space in my company's SAN? If so, great!
My point it that using more disk space has more costs than just the price of a drive platter.
eric
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
Re: Request to re-open MNG-3472
Posted by Ron Wheeler <rw...@artifact-software.com>.
On 02/11/2010 1:20 PM, Haszlakiewicz, Eric wrote:
>> -----Original Message-----
>> From: Martin Gainty [mailto:mgainty@hotmail.com]
>> Sent: Friday, October 29, 2010 1:14 PM
>> To: users@maven.apache.org
>> Subject: RE: Request to re-open MNG-3472
>>
>> Hi Brian
>>
>> if diskspace was a concern i would concur but as each repository and their
>> mirrors support terabyte capacity i would hold off on the purge
>> i was bitten by a wizbang version maven plugin that wasnt tested with
>> wizbank injector and failed when the maven-plugin's were injected using
>> incorrect role and roleHint
> What in the world are you talking about? How can you claim that everyone's has the capability and the money to buy terabytes of disk space, especially for everyone's local repositories?
>
>>> From: Brian.Levine@nokia.com
>>> To: users@maven.apache.org
>>> Date: Fri, 29 Oct 2010 15:28:27 +0200
>>> Subject: RE: Request to re-open MNG-3472
>>>
>>> I'm not suggesting that Maven periodically run a task to purge the repo.
>> I'm suggesting that a settable policy be implemented such that a max number
>> of versions of a snapshot artifact be kept in the local repo. When that
>> limit is exceeded, the oldest version is deleted.
>>> Yes, it's easy enough to do an rm -rf on your local repo. But that
>> assumes that you understand why you need to do this or have been told by
>> your dev team to do this and that you remember to do it before you realize
>> that a lot of disk space is being used up by something.
> This is even more of a problem when you have dozens of people performing builds on a variety of shared machines. It seems like many people don't realize that not everyone limits their use of maven to personal PCs with huge disks. When you have several people working on the same system "oh, it's just a few GB" quickly ends up turning into "out of disk space".
>
The guys with small disk can just delete their entire local repo and let
maven rebuild it by itself from your central server which should have
lots of space.
One or two builds usually gets us back to a fast build from localhost.
If your central server is short of space spend $100 and add a terabyte or 2.
>>> IMHO, a tool should never have unbounded access to resources such as disk
>> space. Further, I don't think stakeholders (not all of which are
>> developers) should need to understand the inner working of Maven or what
>> they need to do work around a limitation such as this.
>>> -brian
> I definitely agree with this. It's hard enough to get people to clean up the files that the explicitly create.
>
> eric
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
> For additional commands, e-mail: users-help@maven.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org