-
-
Notifications
You must be signed in to change notification settings - Fork 404
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
When boost is running, graphs can appear broken #4941
Comments
FOund this when running boost manually php poller_boost.php --verbose --debug --force >> /tmp/boost.txt
|
Output from debug file
|
Boost tables are clean according to audit_database bash-4.2$ php audit_database.php --report | grep boost |
OK, so when I ran boost the first time manually I think I collided with cacti running it? |
Viewing graphs can break when boost is running in some rare cases
Test now @bmfmancini |
So far so good @TheWitness will let it soak for a bit and let you know |
@TheWitness unfortunately still seeing gaps in plotting |
confirmed its only after a graph has been viewed and boost runs afterwards |
Any errors in the log? |
No error at all
…On Fri., Oct. 7, 2022, 5:09 p.m. TheWitness, ***@***.***> wrote:
Any errors in the log?
—
Reply to this email directly, view it on GitHub
<#4941 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/ADGEXTGBFSJ5UXRHI25BUHLWCCGPNANCNFSM6AAAAAAQ32N42A>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Well, that's good. Now you have to find the real reason. How many poller items for the device in question? |
Various amounts of poller items and different templates
Alot of them have no commonality device type wise and it at random times
but always seems to be associated with a boost run
…On Sat., Oct. 8, 2022, 11:56 a.m. TheWitness, ***@***.***> wrote:
Well, that's good. Now you have to find the real reason. How many poller
items for the device in question?
—
Reply to this email directly, view it on GitHub
<#4941 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/ADGEXTBHWD6N5HE26L3Q2M3WCGKRZANCNFSM6AAAAAAQ32N42A>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
You need to very specific. If there is more than one device, give me a count for each. |
Ok will do
…On Sat., Oct. 8, 2022, 12:33 p.m. TheWitness, ***@***.***> wrote:
You need to very specific. If there is more than one device, give me a
count for each.
—
Reply to this email directly, view it on GitHub
<#4941 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/ADGEXTH2K3YKZOFHPZE7MO3WCGO55ANCNFSM6AAAAAAQ32N42A>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
What RRDtool version? |
I have another theory... |
However, you need to answer the poller items question for a few of the cases. |
Rrd version is 1.4
…On Mon., Oct. 10, 2022, 5:55 a.m. TheWitness, ***@***.***> wrote:
However, you need to answer the poller items question for a few of the
cases.
—
Reply to this email directly, view it on GitHub
<#4941 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/ADGEXTGAQXGZ57FOTJDIRGDWCPR2PANCNFSM6AAAAAAQ32N42A>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Upgrade to 1.8 |
Ok updated to rrdtool 1.8
Gaps are still being seen after viewing a graph the data for that time period is removed from the poller_output_boost table while checking for data the poller_output_boost table will have entries in it for that data source and they will disappear from the table while the graph still shows a gap however on the next boost run the graph will start to plot again but only with the data that populated in the table since its been viewed Here are my steps 1.) View poller_output_boost table
2.) View the graph 3.) Check poller_output_boost table entries will be removed for the timespan you are viewing except for new polled data
Graph will still show the gap until boost run but only the newly polled data will make it to the rra |
oops forgot to add the poller count for this example device is 12 |
@TheWitness is the above info all what you were looking for ? |
This change simplifies the function to handle only a single local data id since the mass boost update uses it's own function.
Okay, bug was confirmed and this is resolved now. |
This is still broken when boost is running. Looking to get a fix together. |
The previous sort algo was not sorting properly. Replacing with one that is known good.
@bmfmancini I'm going to mark this resolved. If after updating tomorrow to the latest in test, you find issues, we can re-open. |
So, I had to back out this version of lib/boost.php today due to some sort a new failure, but I have been unable to reproduce in my lab. Still looking into it. This is a minor change that I'm making as a part of my bug hunt.
Okay, working now. Type and no log.
Hey All,
I have found that on 1.2.22 in between boost runs some devices show gaps in the graph
I traced the data from spine all the way to poller_output_boost
the data makes it way correctly until it hits the RRA write it seems that for some the data is not written
The data is removed out of poller_output_boost after each run so its not stuck in there
No relevant errors in log
For the affected devices I have rebuilt the poller cache no difference
The text was updated successfully, but these errors were encountered: