GLBSE down?

[mila]

I see it on and off again. Mostly off (unreachable).

[1714968797]

1714968797

[1714968797]

1714968797

[1714968797]

1714968797

[1714968797]

1714968797

[1714968797]

1714968797

[1714968797]

1714968797

[teek]

Anyone know why charts.glbse.com has been flaking alot lately?

[molecular]

it's been a while, but it seems glbse is down again ;|

[marked]

it's been a while, but it seems glbse is down again ;|

It's getting ridiculous, one of the mining co.s pays divs Friday evening/Sat morning, and the site goes down, comes up, next mining co, down, repeat until Monday maybe Tuesday.

I do hope GLBSE 2.0 has database and bitcoind monitoring - it can't be that difficult to have a script that periodically polls to notice whether either is down and restart them.

But then again maybe the only way is if someone PMs nefario.


marked, (puts away sledgehammer)

[molecular]

http://www.pingdom.com/

will this help him? ill pay the 10$ monthly fee for him if it will :/

I'm sure he has the means to do this himself. Don't know why he doesn't... maybe he did and it's not working fully.

[mila]

is up.

[marked]

http://www.pingdom.com/

will this help him? ill pay the 10$ monthly fee for him if it will :/

Wouldn't help - the web server and pages are generally always available. It is the backend (bitcoind and/or database) that seems to fall over.

Because of the nature of the method used to access the accounts it would be difficult to verify using standard monitoring/logging tools. They generally use auth mechanisms that are standard over http or https. It would require a custom plugin that uses the bmc commandline tools to use.

Alternatively you could follow the GLBSE twitter feed and if no trades have occurred in 30 mins/1hour see if it is down. Or even have a dummy account and tradeable item that is sold and bought every couple of minutes, which isn't visible to general users.


marked

[mila]

... have a dummy account and tradeable item that is sold and bought every couple of minutes, which isn't visible to general users.

unless nefario fixes 0 volume trades you can test by simply selling anything for .01 and pollute a bit the twitter stream and that's asset trade history

[marked]

unless nefario fixes 0 volume trades you can test by simply selling anything for .01 and pollute a bit the twitter stream and that's asset trade history

could just be a private twitter stream, alternatively buying and selling a 1 share asset.

marked

[mila]

unless nefario fixes 0 volume trades you can test by simply selling anything for .01 and pollute a bit the twitter stream and that's asset trade history

could just be a private twitter stream, alternatively buying and selling a 1 share asset.

I got your point of having a dedicated QA asset that could just be 'not public' and use that for trading / testing of glbse.
i want to avoid that because once we have play assets, we can also have play money within glbse and once that money leaks and ends up circulating for real assets, we've invented fractional reserves. the less fake assets the better. fake money should be taboo

[marked]

I got your point of having a dedicated QA asset that could just be 'not public' and use that for trading / testing of glbse.
i want to avoid that because once we have play assets, we can also have play money within glbse and once that money leaks and ends up circulating for real assets, we've invented fractional reserves. the less fake assets the better. fake money should be taboo

Reasonable point.

It could be just a 1 share for an asset value of 1 satoshi. (0.00000001BTC). It is just something that could be implemented immediately for little work.

Thinking about it a little more it would require an ability to limit trades to certain User IDs as well which I don't know if GLBSE supports?

marked

[mila]

It could be just a 1 share for an asset value of 1 satoshi. (0.00000001BTC). It is just something that could be implemented immediately for little work.

Thinking about it a little more it would require an ability to limit trades to certain User IDs as well which I don't know if GLBSE supports?

if we get asset with ask .0000002 and bid .00000001 total 1000 issues we could sell back and forth with any user even w/o additional code.
wait, someone could buy up all the stock and screw the test script ... it could work however if we behave and/or the test script would check if there is open sell order an release ~5 shares each hour for testing ... if we keep talking satoshi valued stock, should be no problem to fund such tests with real bitcoin and follow official rules. just maybe tweak the twitter stream not to publish this asset. anyway it would be probably a bit of working (and maybe twitter QA msgs would be appreciated if it's reasonable amount per day, like only make a trade if there was no activity last 60 minutes; assuming such a script would be run hourly)

[marked]

if we get asset with ask .0000002 and bid .00000001 total 1000 issues we could sell back and forth with any user even w/o additional code.

er no, that asset issues 1 share only, traded at 1 satoshi. Therefore only one can be sold, and thus bought. No need for any more or for it to be public. It can only be sold/bought by 2 userids.

Actually I remembered you can transfer assets, which I don't think shows on the twitter stream. Do frozen assets show as anything different? and can they be transferred (I haven't seen what if any difference occurs on v1 against 2.0) That should be better all round, without the messing of buying / selling. Two independently running scripts monitoring the two accounts, to transfer and validate alternately. The script should be run at least every minute, 5 minutes at most.

marked

[mila]

actually transfers work fine, thus if user A can transfer to user B glbse is up
and we don't need the QA sathoshi valued asset and limitations around it
who can and who can't trade it : ) if test would be based on transfer use case

[Nefario]

Watching 

[mila]

Watching 

the most innovative approach to monitor own site's uptime
one more reason to use glbse

[Nefario]

Being perfectly honest, this is one of the reasons I'm working on GLBSE 2.0, to get rid of this problem. And on that note, I'm almost ready to put up the latest changes to dev.glbse.com, which has basically just be user input validation.

[mila]

Being perfectly honest, this is one of the reasons I'm working on GLBSE 2.0, to get rid of this problem. And on that note, I'm almost ready to put up the latest changes to dev.glbse.com, which has basically just be user input validation.

thumbs up for improvements on that (input validation), is badly needed

while whining over the soon to be gone design of the 1.0 web client I was thinking if the API would be powerful enough to allow custom clients and the reimplementation of the current design (or something of comparably elegant and simple) as I can't get used to the new look & feel. let's see what the dev version brings and post a handful of new bugs found (I know at leas one more dedicated user who's willing to sync and perform coordinated testing once you release something new).

would you mind to indicate & update the version already since software is in this pre-release state? it would help the guerrilla testers to better reference bugs' affect/fix versions and allow better focus on what needs to be retested between the site updates. without release notes it's hard to track versions of what is to be tested. it would also help to know where you version the code and maybe add some lyrics //comments to your code if it's going to be public. not volunteering developing but willing to help with the release mgmt folklore and QA in general.
either as a date/timestamp of the code that you install in glbse or incrementing in old fashioned way 2.01, 2.02 etc.

the uptime checking of glbse 1.0 would not be that difficult. I strongly humbly recommend using it as long as possible. because
1) i like it more
2) it has great record of security (not yet been compromised afaik, surely not only due to the epic downtimes but also because of the quality of the current code, and the use of webkeys -  should be pretty hard to brute force)
3) has only minor defects which do not affect functionality (create buy or sell that is paired immediately and it won't display list of tx that result from that closing; one has them the other not, not sure which one is which. history in web client is limited to 10 entries; the python client can browse full history. some history pages sort newest first, some sort the other way.)

[Nefario]

surely not only due to the epic downtimes

GLBSE's greatest strength, our epic downtime.

We'll I think v1 still has some merit, and I'd like to be able to continue to support it once 2.0 is out, but I simply don't have the resources, my time is stretched already as it is.

What I will do is release GLBSE 1.0 as opensource, then you get to laugh at my code as well as the downtime

[mila]

We'll I think v1 still has some merit, and I'd like to be able to continue to support it once 2.0 is out, but I simply don't have the resources, my time is stretched already as it is.

I'd like to remind you that a day has 24 hours and if you cant finish your tasks during the day, work at night as well


I read this

What I will do is release GLBSE 1.0 as opensource, then you get to laugh at my code as well as the downtime

as 2.0 is closed source, 1.0 is closed source but will be open sourced.

would you also indicate (at least) release date as datum somewhere on the dev.glbse site? for more accurate tracking of bugs and better knowing when new testing is due.