Eventlet before 0.35.2, as used in dnspython before 2.6.0, allows remote attackers to interfere with DNS name resolution by quickly sending an invalid packet from the expected IP address and source port, aka a "TuDoor" attack. In other words, dnspython does not have the preferred behavior in which the DNS name resolution algorithm would proceed, within the full time window, in order to wait for a valid packet. 
NOTE: dnspython 2.6.0 is unusable for a different reason that was addressed in 2.6.1.

Eventlet 0.34.3 addresses a security concern by adhering to RFC 9112 standards, now rejecting requests that include both content-length and transfer-encoding headers simultaneously, a practice that violates the RFC and could pose security risks. However, the reject_bad_request option, if disabled, allows these requests for compatibility with older servers that cannot be updated.
https://github.com/eventlet/eventlet/pull/826/commits/3e50a26878ed08e6a7f510d8d3bc49cf7ee970ed

Eventlet affected versions contain a vulnerability in HTTP header processing within the WSGI implementation. This flaw could allow attackers to craft specially formatted headers, potentially causing header name collisions or overwriting important headers in the WSGI environment. Such manipulation could lead to security issues depending on how applications process these headers.

Affected versions of the eventlet package are vulnerable to Request Smuggling due to improper handling of HTTP trailer sections. The eventlet.wsgi module’s _chunked_read method read only a single line after the terminating 0\r\n chunk and failed to consume remaining trailer lines, leaving bytes in the connection buffer that could be interpreted as the start of a subsequent request.

Eventlet 0.31.0 includes a fix for CVE-2021-21419: A websocket peer may exhaust memory on Eventlet side by sending very large websocket frames. Malicious peer may exhaust memory on Eventlet side by sending highly compressed data frame. A patch in version 0.31.0 restricts websocket frame to reasonable limits. As a workaround, restricting memory usage via OS limits would help against overall machine exhaustion, but there is no workaround to protect Eventlet process.

Eventlet before 0.35.2, as used in dnspython before 2.6.0, allows remote attackers to interfere with DNS name resolution by quickly sending an invalid packet from the expected IP address and source port, aka a "TuDoor" attack. In other words, dnspython does not have the preferred behavior in which the DNS name resolution algorithm would proceed, within the full time window, in order to wait for a valid packet. 
NOTE: dnspython 2.6.0 is unusable for a different reason that was addressed in 2.6.1.

Eventlet 0.34.3 addresses a security concern by adhering to RFC 9112 standards, now rejecting requests that include both content-length and transfer-encoding headers simultaneously, a practice that violates the RFC and could pose security risks. However, the reject_bad_request option, if disabled, allows these requests for compatibility with older servers that cannot be updated.
https://github.com/eventlet/eventlet/pull/826/commits/3e50a26878ed08e6a7f510d8d3bc49cf7ee970ed

Eventlet affected versions contain a vulnerability in HTTP header processing within the WSGI implementation. This flaw could allow attackers to craft specially formatted headers, potentially causing header name collisions or overwriting important headers in the WSGI environment. Such manipulation could lead to security issues depending on how applications process these headers.

Affected versions of the eventlet package are vulnerable to Request Smuggling due to improper handling of HTTP trailer sections. The eventlet.wsgi module’s _chunked_read method read only a single line after the terminating 0\r\n chunk and failed to consume remaining trailer lines, leaving bytes in the connection buffer that could be interpreted as the start of a subsequent request.

Eventlet 0.31.0 includes a fix for CVE-2021-21419: A websocket peer may exhaust memory on Eventlet side by sending very large websocket frames. Malicious peer may exhaust memory on Eventlet side by sending highly compressed data frame. A patch in version 0.31.0 restricts websocket frame to reasonable limits. As a workaround, restricting memory usage via OS limits would help against overall machine exhaustion, but there is no workaround to protect Eventlet process.