https://github.com/osquery/osquery logo
Join Slack
Powered by
Quick question, is the numerical ID hosts are iden...
# kolide
t
Quick question, is the numerical ID hosts are identified with in Kolide (the one needed to be sent to the API to do a delete command) supposed to increment by 1 when new hosts are added?
z
yes
t
ok, I have some systems that have normalish numbers like 28, 57 and so on. After doing some testing with sending Delete commands and accidentally sending some high number host IDs, my new host IDs are in the 100,000 range. I've never had this many hosts connect and was trying to figure out why the ID is so high now.
z
I’m not sure this will have any negative effects, but I am curious to explore. Can you document what you know in a GitHub issue please?
t
Sorry for the delay but I finally just opened an issue for this https://github.com/kolide/fleet/issues/2059
Hey Zach I'm looking for a quick fix to reset my host ID issue. In theory would deleting the "hosts" table from my DB force kolide to make a new table and start the host ID back to 1?
z
Hey Tim, I would not advise doing that. Fleet will not automatically recreate the table. It's not clear to me why you're seeing the behavior you are, and I'm not able to reproduce it. Do you believe it is causing problems for you?
t
Right now it is not causing any problems but my highest ID is 7530599 right now and I am going to have thousands more hosts checking in within a few weeks. From what I saw in the DB the highest number it can handle is 4294967295. If Fleet won't recreate the table would truncating the table work? Since osqueryd is running on all the hosts with the appropriate flags file I'm thinking they would all just check in again and get a new ID. Also, I've been looking into the DB deeper and noticed the big jump in IDs happens with each new date. Here's some of my hosts table with IDs and created_at.
Copy code
mysql> select id,created_at from hosts count ;
+---------+---------------------+
| id      | created_at          |
+---------+---------------------+
|       9 | 2019-04-12 22:19:06 |
|     171 | 2019-05-09 17:39:30 |
|     224 | 2019-05-09 17:42:14 |
|     226 | 2019-05-09 17:42:14 |
|     228 | 2019-05-09 17:42:16 |
|     251 | 2019-05-09 17:43:22 |
|     254 | 2019-05-09 17:43:25 |
|     264 | 2019-05-09 17:43:34 |
|     794 | 2019-05-09 18:11:26 |
|   33727 | 2019-05-10 15:14:35 |
|   33734 | 2019-05-10 15:14:46 |
|   33735 | 2019-05-10 15:14:46 |
|   33742 | 2019-05-10 15:14:55 |
|  143650 | 2019-05-13 13:09:00 |
|  171565 | 2019-05-13 19:02:03 |
z
Are you seeing this kind of jump in any other tables?
t
it doesn't seem to be happening in other tables. When I add a new user their ID increments by one.
turned on some logging for the DB and say this INSERT line show up a ton, which explains the increment for new hosts being much higher than exepected.
Copy code
1561764009535540,kolide-osquery,kolide_user,10.179.135.46,528065,2686921774,QUERY,kolide,'INSERT INTO hosts (    detail_update_time,    osquery_host_id,    seen_time,    node_key   ) VALUES (?, ?, ?, ?)   ON DUPLICATE KEY UPDATE    node_key = VALUES(node_key),    deleted = FALSE',0
Fleet is making a ton of INSERTs and queries constantly, pretty sure this is not intended behavior. I have full log files I can send if you'd like.
so after thinking about what could be causing fleet to be trying to make entries I came across a couple IP's contacting the server and then getting the below error.
Copy code
{
  "component": "http",
  "err": "authentication error: invalid node key: WTf9lH6ZQFe310M49TMlZi+GE/T2fsPS",
  "ts": "2019-06-29T00:54:18.454937871Z"
}
From what I could find in the documentation and your previous conversations in Slack, the node key is assigned to a host by the server after a valid enroll secret is provided. What could be causing the node key to be invalid?
Open in Slack
PreviousNext
Powered by