We are committed to providing responsible disclosure on security and
privacy related incidents. As such, we expect that security or privacy
related incidents are reported directly to us through
<email@example.com> (or another direct
communication channel) and in return we will verify the incident and
report the problem as fast as possible. This means that we will report the
issue (and possible fixes) directly to affected customers so that they
have time to mitigate the issue before we report the issue publicly. For
privacy-related incidents that affect users on our hosted services we will
report the incident directly to the users and to relevant authorities.
As a consequence of the above policy, it may take several days before we make a public disclosure about a problem. We do provide bug bounties for serious security issues. Also, please let us know if you would like to be mentioned by name as the originator of a incident report.
2019 November 04
Fixed in 1.6.7 Introduced in 1.6.2
We were made aware that it was possible for an authenticated user to list all users in a LogScale cluster using the GraphQL API. This also applies for customers on our cloud services, where it would be possible to get a list of user email addresses. We will take measures to avoid similar issues going forward.
2019 April 04
Fixed in 1.5.6
We were made aware of a vulnerability in our LDAP integration, for
self-hosted customers using using LDAP, who have set
ldap with an ldap service allowing
anonymous bind; it is not relevant for customers using the
ldap-search authentication method.
This is a problem for all releases prior to 1.5.6.
Users logging in with an empty password with such a configuration would be let in without further authentication which is a surprising side effect of the default configuration for some ldap products (see description at LDAP Operation.
Operators can search for
#type=humio loglevel=INFO class=*LdapBindLocalLogin @rawstring="*=0"in the humio repository to identify potential breaches.
No customers on our cloud services were affected by this bug. Initially we only disclosed this to the customers we knew were utilizing a configuration where this could be a problem.
2019 March 25
Fixed in 1.5.2
We were made aware of a vulnerability in the authentication for the audit-log repo allowing users to access data for other users when querying using the /query API.
This is a serious bug which could potentially have leaked information about our customers; we were however able to verify that no customers on our cloud services were affected by this bug.
2018 December 20 Fixed 2018-12-20
We received a notification that our demo-site
demo.humio.com exposed the email
addresses of other users on the website. This was available in the
settings -> members page for the GitHub data repository on
demo.humio.com. No information
other than email addresses was exposed — not names, IP addresses
or any other information.
This issue is only for the demo website, not for any of our production services.
We only have concrete knowledge of a single incident where someone saw the list (the reporter of the incident, thanks!), but we cannot guarantee that someone else on the list has retrieved and stored it.
We have stopped the service of the demo site, and are fixing the code and configuration so that this does not happen again.
We are truly sorry this happened. Our team is reviewing all of our other processes and procedures to ensure no personal information is exposed elsewhere.
2017 May 14
Sunday night at 22:53 UTC our build server was compromised. An attacker
vulnerability in Jenkins to run a monero (crypto coin) miner.
With the help of the security team at Chainalysis, we have been able to determine, with high confidence, that no information was leaked as a consequence of the server being compromised.
To repeat: we have been able to determine that it is highly unlikely that any information has been leaked. This is the case for both information relating to our LogScale Cloud customers and our self-hosted customers. Likewise, none of the releases available to customers have been affected. The vulnerability was, as far as we have been able to determine, only exploited to run the said mining program to the effect that our build server became so slow that we were unable to do a build.
While doing this security investigation however, we have found some
things we would like to improve, and thus we're now in the process of
rebuilding our infrastructure from scratch. As a result, some LogScale
Cloud customers can expect observe short service outages on the order of
~5 minutes over the next few days
as our we update DNS to a rebuilt production environment.
In the process we have also found the attacker's command-and-control server, which also lists the ~2700 other hosts that have been attacked in a similar way. We are working on the best way to provide information to those unfortunate fellows.