curl, or an application that uses libcurl, may have a problem with an SSL certificate that works fine when using a web browser to access the same URL. Typical error output from curl looks like this:

$ curl -v https://my-subdomain.mysecuresite.com

 Trying xxx.xxx.xxx.xxx:443…
 TCP_NODELAY set
 Connected to my-subdomain.mysecuresite.com (xxx.xxx.xxx.xxx) port 443 (#0)
 ALPN, offering h2
 ALPN, offering http/1.1
 successfully set certificate verify locations:
 CAfile: /etc/ssl/certs/ca-certificates.crt
 CApath: none
 TLSv1.3 (OUT), TLS handshake, Client hello (1):
 TLSv1.3 (IN), TLS handshake, Server hello (2):
 TLSv1.2 (IN), TLS handshake, Certificate (11):
 TLSv1.2 (OUT), TLS alert, unknown CA (560):
 SSL certificate problem: unable to get local issuer certificate
 Closing connection 0
 curl: (60) SSL certificate problem: unable to get local issuer certificate
 More details here: https://curl.haxx.se/docs/sslcerts.html 
 curl failed to verify the legitimacy of the server and therefore could not
 establish a secure connection to it. To learn more about this situation and
 how to fix it, please visit the web page mentioned above.

Troubleshooting Strategy

1. A good starting point for any SSL error on a public-facing URL is to analyze the URL at SSL Labs.
2. Does the error happen for all certificates issued by a specific Certificate Authority (CA)? If so, the system running curl may need to have a root certificate for that CA added or updated in its certificate repository. That’s a relatively rare problem, but might occur if the system running curl is very old. There’s a comprehensive thread about this issue on Stack Overflow.
3. If the error only happens for one specific site, it’s likely that the site is missing an intermediate certificate. The command to diagnose this issue is also found in that Stack Overflow thread:

openssl s_client -connect myhost.com:443 -servername myhost.com -showcerts

The output should show a series of certificates, starting with the site certificate, and ending with the root certificate for the Certification Authority. If this chain only shows the site certificate, that’s the problem.

Ubiquiti publishes detailed data sheets for all of their products, but it’s hard to find all the information you need in one place for comparison purposes. That’s the purpose of this article.

First Generation Access Points

It can be confusing to distinguish between the generations of UniFi access points; you have to look at the access point label, or access it via the UniFi software to see the version. The performance difference between generations are significant. The datasheet for first generation Ubiquiti UAP wireless access points is still available.

Most business-class switches and routers support monitoring via SNMP. The main barrier to implementing SNMP monitoring is that so many parameters can be monitored, and it can be very hard to isolate the most important parameters. This example shows how to monitor the status of a specific port. Note that the term “port” includes link aggregation groups (LAGs) or other logical groups of ports like VLANs.

Use Case

A top-of-rack switch has several important link aggregation groups (LAGs). For example, one LAG has connections to two identical routers in a VRRP cluster. Other LAGs provide redundant network paths to independent NICs on the same server. I want to be alerted if the port status changes on any critical LAG. The ISP delivers Internet to a single port on the switch, which I also want to monitor.

openssl is the standard open-source, command-line tool for manipulating SSL/TLS certificates on Linux, MacOS, and other UNIX-like systems. I recently ran into an interesting problem using openssl to convert a private key obtained from GoDaddy. Someone else used GoDaddy’s “wizard” interface to generate a certificate signing request (CSR) and private key, and saved the files on their Windows workstation. They purchased an SSL cert from GoDaddy, and shared all the files with me for installation on servers. GoDaddy saved the private key in the newer PKCS #8 format (pkcs8), and one system required the key in the older PKCS #1 (pkcs1) format. It’s easy to tell the difference.

The latest Nginx rpm (nginx-1.16.1-1.el7) from EPEL for CentOS/Redhat is a fairly major update, from version 1.12 to 1.16, and includes some very important fixes to address multiple CVEs. There is another small but critical change in the latest RPM that isn’t mentioned in the bug report. RedHat had configured nginx to run as user “apache” in past releases, but in the EPEL release on 19 October 2019, the configuration was changed to run as user “nginx”. If you ran “yum upgrade” recently (which you really need to do on a regular basis) on any system with nginx installed, you probably have a system that is affected.

PERMISSION DENIED (Photo by Hidde van Esch on Unsplash)

If you recently updated your RedHat or CentOS 7.6 system, you may suddenly start getting “Permission Denied” errors when attempting to mount SMB shares via CIFS. Typical error messages in syslog look like this:

kernel: Status code returned 0xc000006d STATUS_LOGON_FAILURE
kernel: CIFS VFS: Send error in SessSetup = -13
kernel: CIFS VFS: cifs_mount failed w/return code = -13

Configuration that triggered the problem

Synology NAS with latest operating system shares a volume via SMBCentOS 7 Linux server mounts SMB share using a local username and password (NOT domain credentials)cifs-utils version 6.2 installed in May of 2019 (this by itself worked fine)libsmbclient just updated to 4.9.1-6libmount just updated to 2.23.2-61CIFS mount options: vers=3.0,credentials=/root/credentials.txt,sec=ntlmsspi* File /root/credentials.txt contained a username and password that are LOCAL to the SMB server

You probably have at least a few business-class switches and routers silently do their job, year after year. Network devices can be so reliable that we configure them once, and then forget that they exist. However, forgetting them is a really bad idea, since they require regular OS or firmware updates to patch vulnerabilities, and at some point, they will become unreliable and fail. If your device fails, do you have the configuration available to apply to its replacement? Are you sure the configuration that’s in your archive or documentation is up to date? This post deals with automating backups from Netgear switches and Ubiquiti EdgeRouters (and possibly switches, if they run EdgeOS). Please check out the accompanying repo at https://github.com/craigafinch/network-backup-scripts

Google Cloud SQL recently introduced the capability for an instance to have an IP address within the private address space of your VPC network. Previously, all Cloud SQL instances had IP addresses within the address space of the public Internet. Obviously, the public IP was a security and privacy concern, as well as a potential violation of some compliance and governance frameworks. Now, each instance may have a public IP, private IP, or both. There’s another potential benefit to using a private IP address: performance!

I have a QNAP NAS at home which I use for file storage. It shares several folders with the family’s Mac, Linux, and Windows computers. Many of these files are are irreplaceable, especially our digital photos, so I back them up to the cloud. QNAP provides the Hybrid Backup and Sync application, which supports multiple cloud storage endpoints. QNAP also provides specific documentation for backing up and syncing to Google Cloud Storage. However, QNAP’s instructions don’t explain how to set up a Google Cloud Storage bucket destination, so here is the procedure I used to set up Google Cloud Storage to back up my QNAP NAS.